Ask AI

OpenAI Models

AvalAI provides access to a wide range of OpenAI models through our unified API. This page details the available OpenAI models, their capabilities, and optimal use cases, focusing on the latest generations.

Chat, Reasoning, and Multimodal Models

GPT-5.6 Family

OpenAI's GPT-5.6 family brings three new flagship-grade models to AvalAI: Sol for maximum capability, Terra for balanced everyday production work, and Luna for cost-efficient high-volume reasoning. The family is designed for agentic coding, long-horizon knowledge work, tool-heavy workflows, professional document analysis, and scientific reasoning while improving performance per dollar.

Feature	Details
Model IDs	`gpt-5.6-sol`, `gpt-5.6-terra`, `gpt-5.6-luna`
Context window	1,000,000 input tokens, 128,000 output tokens
Input modalities	Text, Image
Output modalities	Text
Supported endpoints	`v1/chat/completions`, `v1/responses`, `v1/messages`
Reasoning effort	none (default), low, medium, high, xhigh, max where supported
Strengths	Agentic coding, long-context reasoning, professional knowledge work, design judgment, tool orchestration
Best for	Complex coding agents, research assistants, document workflows, production support, cost-tiered model routing

Model	Input pricing	Cached input pricing	Output pricing	Best fit
`gpt-5.6-sol`	$5.00 / 1M tokens	$0.50 / 1M tokens	$30.00 / 1M tokens	Highest-capability work, difficult coding, expert analysis
`gpt-5.6-terra`	$2.50 / 1M tokens	$0.25 / 1M tokens	$15.00 / 1M tokens	Balanced production reasoning and agentic workflows
`gpt-5.6-luna`	$1.00 / 1M tokens	$0.10 / 1M tokens	$6.00 / 1M tokens	Cost-efficient everyday work and high-volume tasks

Key Features:

Three performance tiers: Sol maximizes capability, Terra balances quality and cost, and Luna keeps strong reasoning available at lower cost.
Agentic coding and tool use: Designed for multi-step coding, terminal-style workflows, tool coordination, and long-running professional tasks.
Knowledge work and design judgment: Stronger document, presentation, spreadsheet, and interface refinement capabilities for end-to-end work.
Responses and Messages availability: Use the same model IDs across Chat Completions, Responses, and Messages-compatible workflows where enabled.

python

response = client.chat.completions.create(
    model="gpt-5.6-sol",
    messages=[
        {"role": "system", "content": "You are an expert AI assistant."},
        {
            "role": "user",
            "content": "Review this architecture and identify the highest-risk assumptions with mitigations.",
        },
    ],
)

Responses API version

Use this version when you need Responses-first state, tools, reasoning controls, or typed output items.

python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AVALAI_API_KEY"],
    base_url="https://api.avalai.ir/v1",
)

response = client.responses.create(
    model="gpt-5.6-sol",
    reasoning={"effort": "high"},
    instructions="Return a concise decision, key risks, and a verification checklist.",
    input="Review this architecture proposal for production readiness.",
)

print(response.output_text)

GPT-5.5

GPT-5.5 is OpenAI's newest flagship model, designed for real work across complex, multi-step tasks. It excels at agentic coding, knowledge work, computer use, and scientific research, delivering state-of-the-art performance while matching GPT-5.4 per-token latency.

Feature	Details
Context window	1,000,000 input tokens, 128,000 output tokens
Input pricing	$5.00 / 1M tokens
Cached input pricing	$0.50 / 1M tokens
Output pricing	$30.00 / 1M tokens
Input modalities	Text, Image
Output modalities	Text
Supported endpoints	`v1/chat/completions`, `v1/responses`
Reasoning effort	none (default), low, medium, high, xhigh
Strengths	State-of-the-art agentic coding, long-horizon reasoning, tool use, token efficiency
Best for	Complex coding, agentic workflows, knowledge work, computer use, scientific research

Key Features:

Frontier Performance: State-of-the-art on Terminal-Bench 2.0 (82.7%), Expert-SWE (73.1%), GDPval (84.9%), OSWorld-Verified (78.7%), and Tau2-bench Telecom (98.0%)
Agentic Coding Excellence: Stronger conceptual clarity, long-horizon planning, and reliable tool use across large codebases
Token Efficiency: Uses significantly fewer tokens than GPT-5.4 to complete the same Codex tasks
1M Context Window: Extended input context for reasoning across massive documents and codebases
Full Tool Support: Function calling, structured outputs, web search, file search, image generation, computer use, MCP

python

response = client.chat.completions.create(
    model="gpt-5.5",
    messages=[
        {"role": "system", "content": "You are an expert AI assistant."},
        {
            "role": "user",
            "content": "Refactor this function into an idiomatic async version and explain the trade-offs.",
        },
    ],
)

Responses API version

Use this version when the selected model supports /v1/responses. messages moves to input, and the final text is read from response.output_text.

python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AVALAI_API_KEY"],
    base_url="https://api.avalai.ir/v1",
)

response = client.responses.create(
    model="gpt-5.5",
    instructions="You are a helpful assistant.",
    input="Refactor this function into an idiomatic async version and explain the trade-offs.",
)

print(response.output_text)

messages → input
system message → instructions or a developer item
choices[0].message.content → response.output_text
for tools and multimodal output, inspect response.output by item type.

GPT-5.4 Pro

GPT-5.4 Pro is OpenAI's most advanced reasoning model, designed for complex professional work. This model uses more compute to think harder and provide consistently better answers.

Feature	Details
Context window	1,050,000 input tokens, 128,000 output tokens
Training data	Up to August 31, 2025
Input pricing	$30.00 / 1M tokens
Cached input pricing	$0.30 / 1M tokens
Output pricing	$180.00 / 1M tokens
Input modalities	Text, Image
Output modalities	Text
Supported endpoints	`v1/chat/completions`, `v1/responses`
Reasoning effort	medium, high, xhigh
Strengths	Expert-level reasoning, extended thinking, complex problem-solving
Best for	Research, advanced mathematics, complex coding, strategic planning
Availability	Tier 2, 3, 4, 5 users only

Key Features:

Extended Context: 1.05M token context window for handling massive codebases and documents
Reasoning Token Support: Supports medium, high, and xhigh reasoning effort settings
Multi-Turn Interactions: Designed for multi-turn model interactions before responding
Full Tool Support: Web search, file search, image generation, computer use, MCP support

Note: GPT-5.4 Pro is exclusively available through the Responses API (v1/responses) and supports reasoning.effort: medium, high, xhigh. For prompts with >272K input tokens, pricing is 2x input and 1.5x output. Some requests may take several minutes to complete. Use background mode to avoid timeouts.

python

# GPT-5.4 Pro uses the Responses API
response = client.responses.create(
    model="gpt-5.4-pro",  # Or specific snapshot like gpt-5.4-pro-2026-03-05
    input="Design a comprehensive distributed system architecture with fault tolerance and scalability considerations.",
    reasoning={"effort": "high"},
)

GPT-5.4

GPT-5.4 is OpenAI's frontier model for complex professional work, offering best intelligence at scale for agentic, coding, and professional workflows.

Feature	Details
Context window	1,050,000 input tokens, 128,000 output tokens
Training data	Up to August 31, 2025
Input pricing	$2.50 / 1M tokens
Cached input pricing	$0.25 / 1M tokens
Output pricing	$15.00 / 1M tokens
Input modalities	Text, Image
Output modalities	Text
Supported endpoints	`v1/chat/completions`, `v1/responses`
Reasoning effort	none (default), low, medium, high, xhigh
Strengths	Highest reasoning, configurable effort, coding excellence, vision support
Best for	Complex coding, agentic workflows, multimodal tasks, professional knowledge work

Key Features:

Extended Context: 1.05M token context window
Highest Reasoning: Configurable reasoning effort levels
Full Tool Support: Web search, file search, image generation, code interpreter, computer use, MCP
Distillation Support: Can be used for model distillation
Structured Outputs: Full support for structured outputs and function calling

Note: For prompts with >272K input tokens, pricing is 2x input and 1.5x output for the full session.

python

response = client.chat.completions.create(
    model="gpt-5.4",  # Or specific snapshot like gpt-5.4-2026-03-05
    messages=[
        {"role": "system", "content": "You are an expert AI assistant."},
        {
            "role": "user",
            "content": "Design an algorithm to optimize route planning for a delivery service.",
        },
    ],
)

Responses API version

Use this version when the selected model supports /v1/responses. messages moves to input, and the final text is read from response.output_text.

python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AVALAI_API_KEY"],
    base_url="https://api.avalai.ir/v1",
)

response = client.responses.create(
    model="gpt-5.4",
    instructions="You are a helpful assistant.",
    input="Design an algorithm to optimize route planning for a delivery service.",
)

print(response.output_text)

messages → input
system message → instructions or a developer item
choices[0].message.content → response.output_text
for tools and multimodal output, inspect response.output by item type.

GPT-5.4 Mini

GPT-5.4 Mini is OpenAI's fast, cost-efficient model optimized for high-volume tasks with strong reasoning capabilities.

Feature	Details
Context window	400,000 input tokens, 100,000 output tokens
Training data	Up to August 31, 2025
Input pricing	$0.75 / 1M tokens
Cached input pricing	$0.075 / 1M tokens
Output pricing	$4.50 / 1M tokens
Input modalities	Text, Image
Output modalities	Text
Supported endpoints	`v1/chat/completions`
Reasoning effort	none (default), low, medium
Strengths	Fast responses, cost-effective, strong reasoning, high throughput
Best for	High-volume production, cost-sensitive applications, interactive chat

Key Features:

Fast Responses: Optimized for low-latency generation
400K Context: Extended context window for complex tasks
Reasoning Support: Supports none, low, and medium reasoning effort
Vision Support: Can process images as input
Cost-Effective: Ideal for high-volume applications

python

response = client.chat.completions.create(
    model="gpt-5.4-mini",  # Or specific snapshot
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {
            "role": "user",
            "content": "Summarize the key principles of clean code.",
        },
    ],
)

Responses API version

Use this version when the selected model supports /v1/responses. messages moves to input, and the final text is read from response.output_text.

python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AVALAI_API_KEY"],
    base_url="https://api.avalai.ir/v1",
)

response = client.responses.create(
    model="gpt-5.4-mini",
    instructions="You are a helpful assistant.",
    input="Summarize the key principles of clean code.",
)

print(response.output_text)

messages → input
system message → instructions or a developer item
choices[0].message.content → response.output_text
for tools and multimodal output, inspect response.output by item type.

GPT-5.4 Nano

GPT-5.4 Nano is OpenAI's fastest and most affordable model, designed for simple tasks requiring minimal compute.

Feature	Details
Context window	400,000 input tokens, 100,000 output tokens
Training data	Up to August 31, 2025
Input pricing	$0.20 / 1M tokens
Cached input pricing	$0.02 / 1M tokens
Output pricing	$1.25 / 1M tokens
Input modalities	Text, Image
Output modalities	Text
Supported endpoints	`v1/chat/completions`
Reasoning effort	none (default), low
Strengths	Fastest responses, lowest cost, high throughput
Best for	Simple tasks, classification, basic Q&A, high-volume lightweight workloads

Key Features:

Fastest in Class: Optimized for minimal latency
400K Context: Full extended context window support
Lowest Cost: Most affordable GPT-5.4 series model
Vision Support: Can process images as input
High Throughput: Ideal for batch processing

python

response = client.chat.completions.create(
    model="gpt-5.4-nano",  # Or specific snapshot
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {
            "role": "user",
            "content": "What is the capital of France?",
        },
    ],
)

Responses API version

Use this version when the selected model supports /v1/responses. messages moves to input, and the final text is read from response.output_text.

python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AVALAI_API_KEY"],
    base_url="https://api.avalai.ir/v1",
)

response = client.responses.create(
    model="gpt-5.4-nano",
    instructions="You are a helpful assistant.",
    input="What is the capital of France?",
)

print(response.output_text)

messages → input
system message → instructions or a developer item
choices[0].message.content → response.output_text
for tools and multimodal output, inspect response.output by item type.

GPT-5.3 Chat

GPT-5.3 Chat is the GPT-5.3 Instant model currently used in ChatGPT, optimized for chat use cases with high intelligence at a competitive price point.

Feature	Details
Context window	128,000 input tokens, 16,384 output tokens
Training data	Up to August 31, 2025
Input pricing	$1.75 / 1M tokens
Cached input pricing	$0.175 / 1M tokens
Output pricing	$14.00 / 1M tokens
Input modalities	Text, Image
Output modalities	Text
Supported endpoints	`v1/chat/completions`, `v1/responses`
Strengths	High intelligence, chat optimization, function calling, streaming
Best for	Conversational AI, chatbots, interactive applications

Key Features:

High Intelligence: Latest improvements for chat use cases
Vision Support: Can process images as input
Function Calling: Full support for function calling and structured outputs
Streaming: Full streaming support

Note: We recommend GPT-5.2 for general API usage, but GPT-5.3 Chat is ideal for testing the latest ChatGPT improvements for conversational applications.

python

response = client.chat.completions.create(
    model="gpt-5.3-chat",  # Or gpt-5.3-chat-latest
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {
            "role": "user",
            "content": "Explain quantum computing in simple terms.",
        },
    ],
)

Responses API version

Use this version when the selected model supports /v1/responses. messages moves to input, and the final text is read from response.output_text.

python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AVALAI_API_KEY"],
    base_url="https://api.avalai.ir/v1",
)

response = client.responses.create(
    model="gpt-5.3-chat",
    instructions="You are a helpful assistant.",
    input="Explain quantum computing in simple terms.",
)

print(response.output_text)

messages → input
system message → instructions or a developer item
choices[0].message.content → response.output_text
for tools and multimodal output, inspect response.output by item type.

GPT-5.2 Pro

GPT-5.2 Pro is OpenAI's most advanced reasoning model, producing smarter and more precise responses for difficult problems. It sets new state-of-the-art benchmarks across knowledge work, coding, science, and mathematics.

Feature	Details
Context window	400,000 input tokens, 128,000 output tokens
Training data	Up to August 31, 2025
Input pricing	$21.00 / 1M tokens
Output pricing	$168.00 / 1M tokens
Strengths	Expert-level reasoning, extended thinking, complex problem-solving, professional knowledge work
Best for	Research, advanced mathematics, complex coding, strategic planning, scientific research
Availability	Tier 2, 3, 4, 5 users only
API Endpoint	v1/responses only

Note: GPT-5.2 Pro is exclusively available through the Responses API (v1/responses) and supports reasoning.effort: medium, high, xhigh. Some requests may take several minutes to complete. Use background mode to avoid timeouts.

python

# GPT-5.2 Pro uses the Responses API
response = client.responses.create(
    model="gpt-5.2-pro",  # Or specific snapshot like gpt-5.2-pro-2025-12-11
    input="Design a comprehensive distributed system architecture with fault tolerance and scalability considerations.",
    reasoning={"effort": "high"},
)

GPT-5.2

GPT-5.2 is OpenAI's best general-purpose model for coding and agentic tasks across industries with improved general intelligence, instruction following, and multimodality.

Feature	Details
Context window	400,000 input tokens, 128,000 output tokens
Training data	Up to August 31, 2025
Input pricing	$1.75 / 1M tokens
Cached input pricing	$0.175 / 1M tokens
Output pricing	$14.00 / 1M tokens
Strengths	Advanced reasoning, configurable effort, coding excellence, vision support, multimodal I/O
Best for	Complex coding, agentic workflows, multimodal tasks, professional knowledge work

python

response = client.chat.completions.create(
    model="gpt-5.2",  # Or specific snapshot like gpt-5.2-2025-12-11
    messages=[
        {"role": "system", "content": "You are an expert AI assistant."},
        {
            "role": "user",
            "content": "Design an algorithm to optimize route planning for a delivery service.",
        },
    ],
)

Responses API version

Use this version when the selected model supports /v1/responses. messages moves to input, and the final text is read from response.output_text.

python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AVALAI_API_KEY"],
    base_url="https://api.avalai.ir/v1",
)

response = client.responses.create(
    model="gpt-5.2",
    instructions="You are a helpful assistant.",
    input="Design an algorithm to optimize route planning for a delivery service.",
)

print(response.output_text)

messages → input
system message → instructions or a developer item
choices[0].message.content → response.output_text
for tools and multimodal output, inspect response.output by item type.

GPT-5.3-Codex

GPT-5.3-Codex is OpenAI's most capable agentic coding model to date. It advances both the frontier coding performance of GPT-5.2-Codex and the reasoning capabilities of GPT-5.2, together in one model, which is also 25% faster. This model was instrumental in creating itself, with the Codex team using early versions to debug its own training and manage deployment.

Feature	Details
Context window	400,000 input tokens, 128,000 output tokens
Training data	Up to August 31, 2025
Input pricing	$1.75 / 1M tokens
Cached input pricing	$0.175 / 1M tokens (90% cost reduction)
Output pricing	$14.00 / 1M tokens
Input modalities	Text, Image
Output modalities	Text
Supported endpoints	`v1/responses`
Capabilities	Streaming, Function Calling, Structured Outputs, Reasoning Tokens
Strengths	State-of-the-art coding, agentic workflows, interactive collaboration
Best for	Agentic coding tasks, web development, large codebase refactoring, research acceleration

Key Features:

State-of-the-Art Performance: Sets new industry highs on SWE-Bench Pro (56.8%) and Terminal-Bench 2.0 (77.3%)
Reasoning Token Support: Supports low, medium, high, and xhigh reasoning effort settings
Extended Context: 400K token context window for handling large codebases
High Output Capacity: Up to 128K tokens output for comprehensive code generation
Vision Support: Can process images as input for visual code understanding
Interactive Collaboration: Provides frequent updates during long-running tasks
Self-Improving: First model that was instrumental in creating itself

python

# GPT-5.3-Codex uses the Responses API
response = client.responses.create(
    model="gpt-5.3-codex",
    input="Refactor this codebase to use dependency injection and implement a comprehensive test suite.",
    reasoning={"effort": "high"},
)

GPT-5.2-Codex

GPT-5.2-Codex is OpenAI's intelligent coding model, optimized for long-horizon, agentic coding tasks in Codex or similar environments.

Feature	Details
Context window	400,000 input tokens, 128,000 output tokens
Training data	Up to August 31, 2025
Input pricing	$1.75 / 1M tokens
Cached input pricing	$0.175 / 1M tokens (90% cost reduction)
Output pricing	$14.00 / 1M tokens
Input modalities	Text, Image
Output modalities	Text
Supported endpoints	`v1/chat/completions`, `v1/responses`, `v1/realtime`
Capabilities	Streaming, Function Calling, Structured Outputs, Reasoning Tokens
Strengths	Long-horizon coding, agentic workflows, autonomous code generation
Best for	Agentic coding tasks, large codebase refactoring, autonomous development agents

Key Features:

Reasoning Token Support: Supports low, medium, high, and xhigh reasoning effort settings
Extended Context: 400K token context window for handling large codebases
High Output Capacity: Up to 128K tokens output for comprehensive code generation
Vision Support: Can process images as input for visual code understanding
Agentic Workflows: Optimized for autonomous coding agents and iterative development

python

response = client.chat.completions.create(
    model="gpt-5.2-codex",
    messages=[
        {"role": "system", "content": "You are an expert coding assistant."},
        {
            "role": "user",
            "content": "Refactor this codebase to use dependency injection and implement a comprehensive test suite.",
        },
    ],
    max_tokens=8192,
    temperature=0.6,
)

Responses API version

Use this version when the selected model supports /v1/responses. messages moves to input, and the final text is read from response.output_text.

python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AVALAI_API_KEY"],
    base_url="https://api.avalai.ir/v1",
)

response = client.responses.create(
    model="gpt-5.2-codex",
    instructions="You are a helpful assistant.",
    input="Refactor this codebase to use dependency injection and implement a comprehensive test suite.",
)

print(response.output_text)

messages → input
system message → instructions or a developer item
choices[0].message.content → response.output_text
for tools and multimodal output, inspect response.output by item type.

GPT-5 Pro

GPT-5 Pro is OpenAI's most advanced reasoning model with extended thinking capabilities, designed for complex, expert-level tasks.

Feature	Details
Context window	400,000 input tokens, 272,000 output tokens
Training data	Up to September 30, 2024
Input pricing	$15.00 / 1M tokens
Output pricing	$120.00 / 1M tokens
Strengths	Expert-level reasoning, extended thinking, complex problem-solving
Best for	Research, advanced mathematics, complex coding, strategic planning
Availability	Tier 2, 3, 4, 5 users only
API Endpoint	v1/responses only

Note: GPT-5 Pro is exclusively available through the Responses API (v1/responses) and uses extended reasoning for comprehensive answers. Some requests may take several minutes to complete.

python

# GPT-5 Pro uses the Responses API
response = client.responses.create(
    model="gpt-5-pro",  # Or specific snapshot like gpt-5-pro-2025-10-06
    input="Design a comprehensive distributed system architecture with fault tolerance and scalability considerations.",
)

GPT-5.1

GPT-5.1 is OpenAI's flagship model for coding and agentic tasks with configurable reasoning and non-reasoning effort.

Feature	Details
Context window	400,000 input tokens, 128,000 output tokens
Training data	Up to May 31, 2024
Input pricing	$1.25 / 1M tokens
Cached input pricing	$0.125 / 1M tokens
Output pricing	$10.00 / 1M tokens
Strengths	Advanced reasoning, configurable effort, coding excellence, vision support
Best for	Complex coding, agentic workflows, multimodal tasks, reasoning with images

python

response = client.chat.completions.create(
    model="gpt-5.1",  # Or specific snapshot like gpt-5.1-2025-11-13
    messages=[
        {"role": "system", "content": "You are an expert AI assistant."},
        {
            "role": "user",
            "content": "Design an algorithm to optimize route planning for a delivery service.",
        },
    ],
)

Responses API version

Use this version when the selected model supports /v1/responses. messages moves to input, and the final text is read from response.output_text.

python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AVALAI_API_KEY"],
    base_url="https://api.avalai.ir/v1",
)

response = client.responses.create(
    model="gpt-5.1",
    instructions="You are a helpful assistant.",
    input="Design an algorithm to optimize route planning for a delivery service.",
)

print(response.output_text)

messages → input
system message → instructions or a developer item
choices[0].message.content → response.output_text
for tools and multimodal output, inspect response.output by item type.

GPT-5.1 Chat

GPT-5.1 Chat is a chat-optimized version of GPT-5.1 designed for conversational applications, offering the same advanced capabilities with enhanced optimization for interactive chat experiences.

Feature	Details
Context window	400,000 input tokens, 128,000 output tokens
Training data	Up to May 31, 2024
Input pricing	$1.25 / 1M tokens
Cached input pricing	$0.125 / 1M tokens
Output pricing	$10.00 / 1M tokens
Strengths	Chat optimization, advanced reasoning, coding excellence, vision support
Best for	Interactive conversations, chatbots, conversational AI applications

python

response = client.chat.completions.create(
    model="gpt-5.1-chat",
    messages=[
        {"role": "system", "content": "You are a helpful conversational assistant."},
        {
            "role": "user",
            "content": "Explain the key differences between microservices and monolithic architecture.",
        },
    ],
)

Responses API version

Use this version when the selected model supports /v1/responses. messages moves to input, and the final text is read from response.output_text.

python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AVALAI_API_KEY"],
    base_url="https://api.avalai.ir/v1",
)

response = client.responses.create(
    model="gpt-5.1-chat",
    instructions="You are a helpful assistant.",
    input="Explain the key differences between microservices and monolithic architecture.",
)

print(response.output_text)

messages → input
system message → instructions or a developer item
choices[0].message.content → response.output_text
for tools and multimodal output, inspect response.output by item type.

GPT-5

GPT-5 is OpenAI's flagship model for coding, reasoning, and agentic tasks across domains, representing their most advanced AI capabilities.

Feature	Details
Context window	400,000 input tokens, 128,000 output tokens
Training data	Up to September 30, 2024
Input pricing	$1.25 / 1M tokens
Cached input pricing	$0.125 / 1M tokens
Output pricing	$10.00 / 1M tokens
Strengths	Best-in-class coding, reasoning, agentic tasks, multimodal capabilities
Best for	Complex coding projects, advanced reasoning, agentic workflows, research

python

response = client.chat.completions.create(
    model="gpt-5",  # Or specific snapshot like gpt-5-2025-08-07
    messages=[
        {"role": "system", "content": "You are an expert software architect."},
        {
            "role": "user",
            "content": "Design a scalable microservices architecture for an e-commerce platform.",
        },
    ],
)

Responses API version

Use this version when the selected model supports /v1/responses. messages moves to input, and the final text is read from response.output_text.

python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AVALAI_API_KEY"],
    base_url="https://api.avalai.ir/v1",
)

response = client.responses.create(
    model="gpt-5",
    instructions="You are a helpful assistant.",
    input="Design a scalable microservices architecture for an e-commerce platform.",
)

print(response.output_text)

messages → input
system message → instructions or a developer item
choices[0].message.content → response.output_text
for tools and multimodal output, inspect response.output by item type.

GPT-5 Mini

GPT-5 Mini is a faster, more cost-efficient version of GPT-5, perfect for well-defined tasks with excellent performance.

Feature	Details
Context window	400,000 input tokens, 128,000 output tokens
Training data	Up to May 30, 2024
Input pricing	$0.25 / 1M tokens
Cached input pricing	$0.025 / 1M tokens
Output pricing	$2.00 / 1M tokens
Strengths	Fast, cost-efficient, strong reasoning, good for structured tasks
Best for	Well-defined coding tasks, content generation, analysis with precise prompts

python

response = client.chat.completions.create(
    model="gpt-5-mini",  # Or specific snapshot like gpt-5-mini-2025-08-07
    messages=[
        {"role": "system", "content": "You are a code optimization expert."},
        {
            "role": "user",
            "content": "Optimize this Python function for better performance and readability.",
        },
    ],
)

Responses API version

Use this version when the selected model supports /v1/responses. messages moves to input, and the final text is read from response.output_text.

python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AVALAI_API_KEY"],
    base_url="https://api.avalai.ir/v1",
)

response = client.responses.create(
    model="gpt-5-mini",
    instructions="You are a helpful assistant.",
    input="Optimize this Python function for better performance and readability.",
)

print(response.output_text)

messages → input
system message → instructions or a developer item
choices[0].message.content → response.output_text
for tools and multimodal output, inspect response.output by item type.

GPT-5 Nano

GPT-5 Nano is the fastest, most cost-efficient version of GPT-5, ideal for summarization and classification tasks.

Feature	Details
Context window	400,000 input tokens, 128,000 output tokens
Training data	Up to May 30, 2024
Input pricing	$0.05 / 1M tokens
Cached input pricing	$0.005 / 1M tokens
Output pricing	$0.40 / 1M tokens
Strengths	Very fast, extremely cost-effective, good for simple tasks
Best for	Text summarization, classification, content moderation, high-volume tasks

python

response = client.chat.completions.create(
    model="gpt-5-nano",  # Or specific snapshot like gpt-5-nano-2025-08-07
    messages=[
        {"role": "system", "content": "You are a text classifier."},
        {
            "role": "user",
            "content": "Classify this customer feedback as positive, negative, or neutral: 'The product works well but could be improved.'",
        },
    ],
)

Responses API version

Use this version when the selected model supports /v1/responses. messages moves to input, and the final text is read from response.output_text.

python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AVALAI_API_KEY"],
    base_url="https://api.avalai.ir/v1",
)

response = client.responses.create(
    model="gpt-5-nano",
    instructions="You are a helpful assistant.",
    input="Classify this customer feedback as positive, negative, or neutral:",
)

print(response.output_text)

messages → input
system message → instructions or a developer item
choices[0].message.content → response.output_text
for tools and multimodal output, inspect response.output by item type.

GPT-5 Chat

GPT-5 Chat points to the GPT-5 snapshot currently used in ChatGPT, providing the same experience as the ChatGPT interface.

Feature	Details
Context window	400,000 input tokens, 128,000 output tokens
Training data	Up to September 29, 2024
Input pricing	$1.25 / 1M tokens
Cached input pricing	$0.125 / 1M tokens
Output pricing	$10.00 / 1M tokens
Strengths	ChatGPT-equivalent experience, high intelligence, conversational
Best for	Interactive conversations, general-purpose tasks, ChatGPT-like applications

python

response = client.chat.completions.create(
    model="gpt-5-chat-latest",  # Always points to latest ChatGPT version
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {
            "role": "user",
            "content": "Explain quantum computing concepts in simple terms.",
        },
    ],
)

Responses API version

Use this version when the selected model supports /v1/responses. messages moves to input, and the final text is read from response.output_text.

python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AVALAI_API_KEY"],
    base_url="https://api.avalai.ir/v1",
)

response = client.responses.create(
    model="gpt-5-chat-latest",
    instructions="You are a helpful assistant.",
    input="Explain quantum computing concepts in simple terms.",
)

print(response.output_text)

messages → input
system message → instructions or a developer item
choices[0].message.content → response.output_text
for tools and multimodal output, inspect response.output by item type.

GPT-5 Codex

GPT-5 Codex is a specialized version of GPT-5 optimized for agentic coding tasks, designed specifically for use in Codex environments and advanced programming workflows.

Feature	Details
Context window	400,000 input tokens, 128,000 output tokens
Training data	Up to September 30, 2024
Input pricing	$1.25 / 1M tokens
Cached input pricing	$0.125 / 1M tokens
Output pricing	$10.00 / 1M tokens
Strengths	Optimized for agentic coding, reasoning tokens, advanced programming
Best for	Agentic coding tasks, complex software development, automated programming

python

response = client.chat.completions.create(
    model="gpt-5-codex",
    messages=[
        {"role": "system", "content": "You are an expert programming agent."},
        {
            "role": "user",
            "content": "Create a complete Python class for a binary search tree with insert, search, delete, and traversal methods.",
        },
    ],
)

Responses API version

Use this version when the selected model supports /v1/responses. messages moves to input, and the final text is read from response.output_text.

python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AVALAI_API_KEY"],
    base_url="https://api.avalai.ir/v1",
)

response = client.responses.create(
    model="gpt-5-codex",
    instructions="You are a helpful assistant.",
    input="Create a complete Python class for a binary search tree with insert, search, delete, and traversal methods.",
)

print(response.output_text)

messages → input
system message → instructions or a developer item
choices[0].message.content → response.output_text
for tools and multimodal output, inspect response.output by item type.

GPT-5.1 Codex

GPT-5.1 Codex is the latest advanced reasoning model specifically optimized for agentic coding tasks, debugging, and multi-step programming workflows. This model is only accessible through the v1/responses endpoint.

Feature	Details
Context window	400,000 input tokens, 128,000 output tokens
Training data	Up to May 31, 2024
Input pricing	$1.25 / 1M tokens
Cached input pricing	$0.125 / 1M tokens
Output pricing	$10.00 / 1M tokens
Strengths	Superior agentic coding, advanced debugging, autonomous agent workflows
Best for	Complex code generation, refactoring, autonomous programming agents
Important	Only works with v1/responses endpoint

python

import requests

url = "https://api.avalai.ir/v1/responses"
headers = {
    "Content-Type": "application/json",
    "Authorization": "Bearer $AVALAI_API_KEY",
}

data = {
    "model": "gpt-5.1-codex",
    "messages": [
        {
            "role": "user",
            "content": "Create a complete TypeScript implementation of a thread-safe LRU cache with TTL support.",
        }
    ],
}

response = requests.post(url, headers=headers, json=data)
result = response.json()
print(result["choices"][0]["message"]["content"])

GPT-5.1 Codex Mini

GPT-5.1 Codex Mini is a lightweight version offering excellent code generation capabilities at a more accessible price point. Ideal for high-volume agentic tasks where cost efficiency matters. This model is only accessible through the v1/responses endpoint.

Feature	Details
Context window	400,000 input tokens, 128,000 output tokens
Training data	Up to May 31, 2024
Input pricing	$0.25 / 1M tokens
Cached input pricing	$0.025 / 1M tokens
Output pricing	$2.00 / 1M tokens
Strengths	Cost-efficient coding, fast execution, good for structured tasks
Best for	High-volume coding tasks, automated code reviews, batch processing
Important	Only works with v1/responses endpoint

python

import requests

url = "https://api.avalai.ir/v1/responses"
headers = {
    "Content-Type": "application/json",
    "Authorization": "Bearer $AVALAI_API_KEY",
}

data = {
    "model": "gpt-5.1-codex-mini",
    "messages": [
        {
            "role": "user",
            "content": "Write a Python function to validate email addresses using regex.",
        }
    ],
}

response = requests.post(url, headers=headers, json=data)
result = response.json()
print(result["choices"][0]["message"]["content"])

GPT-5.1-Codex-Max

GPT-5.1-Codex-Max is OpenAI's most intelligent coding model, purpose-built for agentic coding. It is optimized for long-horizon, multi-step programming tasks and autonomous agent workflows.

Feature	Details
Context window	400,000 input tokens, 128,000 output tokens
Training data	Up to September 30, 2024
Input pricing	$1.25 / 1M tokens
Cached input pricing	$0.125 / 1M tokens
Output pricing	$10.00 / 1M tokens
Strengths	Superior agentic coding, long-horizon tasks, advanced reasoning
Best for	Complex multi-step coding, autonomous programming agents, extensive codebase work
Modalities	Text input/output, Image input

python

response = client.chat.completions.create(
    model="gpt-5.1-codex-max",
    messages=[
        {"role": "system", "content": "You are an expert programming agent."},
        {
            "role": "user",
            "content": "Create a complete distributed task queue system with worker management, task prioritization, and failure recovery.",
        },
    ],
)

Responses API version

Use this version when the selected model supports /v1/responses. messages moves to input, and the final text is read from response.output_text.

python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AVALAI_API_KEY"],
    base_url="https://api.avalai.ir/v1",
)

response = client.responses.create(
    model="gpt-5.1-codex-max",
    instructions="You are a helpful assistant.",
    input="Create a complete distributed task queue system with worker management, task prioritization, and failure recovery.",
)

print(response.output_text)

messages → input
system message → instructions or a developer item
choices[0].message.content → response.output_text
for tools and multimodal output, inspect response.output by item type.

o3-pro

o3-pro is OpenAI's most advanced reasoning model for complex problem-solving and analysis, offering the highest level of reasoning capabilities.

Feature	Details
Context window	200,000 input tokens, 100,000 output tokens
Training data	Up to May 31, 2024
Input pricing	$15.00 / 1M tokens
Output pricing	$60.00 / 1M tokens
Strengths	Highest-level reasoning, complex problem-solving, advanced analysis
Best for	Most complex reasoning tasks, professional analysis, research applications

python

response = client.chat.completions.create(
    model="o3-pro",
    messages=[
        {"role": "system", "content": "You are an expert problem solver."},
        {
            "role": "user",
            "content": "Analyze this complex mathematical proof and provide detailed improvements.",
        },
    ],
)

Responses API version

Use this version when the selected model supports /v1/responses. messages moves to input, and the final text is read from response.output_text.

python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AVALAI_API_KEY"],
    base_url="https://api.avalai.ir/v1",
)

response = client.responses.create(
    model="o3-pro",
    instructions="You are a helpful assistant.",
    input="Analyze this complex mathematical proof and provide detailed improvements.",
)

print(response.output_text)

messages → input
system message → instructions or a developer item
choices[0].message.content → response.output_text
for tools and multimodal output, inspect response.output by item type.

o3-deep-research

o3-deep-research is our most powerful deep research model designed to tackle complex, multi-step research tasks. It can search and synthesize information from across the internet as well as from your own data. This model is only accessible through the v1/responses endpoint and requires tools to be selected.

Feature	Details
Context window	200,000 input tokens, 100,000 output tokens
Training data	Up to May 31, 2024
Input pricing	$10.00 / 1M tokens
Cached input pricing	$2.50 / 1M tokens
Output pricing	$40.00 / 1M tokens
Strengths	Deep research capabilities, internet search, multi-step analysis, reasoning tokens
Best for	Complex research tasks, information synthesis, multi-step investigations
Important	Only works with v1/responses endpoint and requires tools

python

import requests

url = "https://api.avalai.ir/v1/responses"
headers = {
    "Content-Type": "application/json",
    "Authorization": "Bearer $AVALAI_API_KEY",
}

data = {
    "model": "o3-deep-research",
    "tools": [{"type": "web_search", "search_context_size": "medium"}],
    "input": "Research the latest developments in quantum computing and their implications",
}

response = requests.post(url, headers=headers, json=data)

Web Search with Domain Filtering

python

import requests

url = "https://api.avalai.ir/v1/responses"
headers = {
    "Content-Type": "application/json",
    "Authorization": "Bearer $AVALAI_API_KEY",
}

data = {
    "model": "o3-deep-research",
    "tools": [
        {
            "type": "web_search",
            "filters": {
                "allowed_domains": [
                    "arxiv.org",
                    "nature.com",
                    "science.org",
                    "ieee.org",
                ]
            },
        }
    ],
    "input": "Research quantum computing breakthroughs from academic sources",
}

response = requests.post(url, headers=headers, json=data)

o3

o3 is OpenAI's most powerful reasoning model, setting a new standard for math, science, coding, and visual reasoning tasks. It excels at technical writing and instruction-following.

Feature	Details
Context window	200,000 input tokens, 100,000 output tokens
Training data	Up to May 31, 2024
Input pricing	$10.00 / 1M tokens
Output pricing	$40.00 / 1M tokens
Strengths	Advanced reasoning, multi-step problems, analysis across text, code, and images
Best for	Complex reasoning tasks, technical problem-solving requiring deep analysis

python

response = client.chat.completions.create(
    model="o3",  # Or specific snapshot like o3-2025-04-16
    messages=[
        {"role": "system", "content": "You are an expert problem solver."},
        {
            "role": "user",
            "content": "Analyze this mathematical proof and identify any errors or improvements.",
        },
    ],
)

Responses API version

Use this version when the selected model supports /v1/responses. messages moves to input, and the final text is read from response.output_text.

python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AVALAI_API_KEY"],
    base_url="https://api.avalai.ir/v1",
)

response = client.responses.create(
    model="o3",
    instructions="You are a helpful assistant.",
    input="Analyze this mathematical proof and identify any errors or improvements.",
)

print(response.output_text)

messages → input
system message → instructions or a developer item
choices[0].message.content → response.output_text
for tools and multimodal output, inspect response.output by item type.

o4-mini-deep-research

o4-mini-deep-research is our faster, more affordable deep research model—ideal for tackling complex, multi-step research tasks. It can search and synthesize information from across the internet as well as from your own data. This model is only accessible through the v1/responses endpoint and requires tools to be selected.

Feature	Details
Context window	200,000 input tokens, 100,000 output tokens
Training data	Up to May 31, 2024
Input pricing	$2.00 / 1M tokens
Cached input pricing	$0.50 / 1M tokens
Output pricing	$8.00 / 1M tokens
Strengths	Fast, affordable deep research, internet search, reasoning tokens
Best for	Cost-effective research tasks, information synthesis, multi-step analysis
Important	Only works with v1/responses endpoint and requires tools

python

import requests

url = "https://api.avalai.ir/v1/responses"
headers = {
    "Content-Type": "application/json",
    "Authorization": "Bearer $AVALAI_API_KEY",
}

data = {
    "model": "o4-mini-deep-research",
    "tools": [{"type": "web_search", "search_context_size": "medium"}],
    "input": "Research current trends in renewable energy technology",
}

response = requests.post(url, headers=headers, json=data)

Web Search with User Location

python

import requests

url = "https://api.avalai.ir/v1/responses"
headers = {
    "Content-Type": "application/json",
    "Authorization": "Bearer $AVALAI_API_KEY",
}

data = {
    "model": "o4-mini-deep-research",
    "tools": [
        {
            "type": "web_search",
            "user_location": {
                "type": "approximate",
                "country": "US",
                "city": "San Francisco",
                "region": "California",
                "timezone": "America/Los_Angeles",
            },
        }
    ],
    "input": "Find renewable energy companies near me",
}

response = requests.post(url, headers=headers, json=data)

o4-mini

o4-mini is OpenAI's latest small o-series model, optimized for fast, effective reasoning with exceptionally efficient performance in coding and visual tasks.

Feature	Details
Context window	200,000 input tokens, 100,000 output tokens
Training data	Up to May 31, 2024
Input pricing	$1.10 / 1M tokens
Output pricing	$4.40 / 1M tokens
Strengths	Fast, cost-effective reasoning, good for visual tasks and coding
Best for	Fast reasoning, coding tasks, visual understanding with cost efficiency

python

response = client.chat.completions.create(
    model="o4-mini",  # Or specific snapshot like o4-mini-2025-04-16
    messages=[
        {"role": "system", "content": "You are a coding assistant."},
        {
            "role": "user",
            "content": "Write a function to check if a string is a palindrome.",
        },
    ],
)

Responses API version This version uses `gpt-5.5` because `o4-mini` may not be enabled for `/v1/responses` in the current AvalAI model data.

Use this version when the selected model supports /v1/responses. messages moves to input, and the final text is read from response.output_text.

python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AVALAI_API_KEY"],
    base_url="https://api.avalai.ir/v1",
)

response = client.responses.create(
    model="gpt-5.5",
    instructions="You are a helpful assistant.",
    input="Write a function to check if a string is a palindrome.",
)

print(response.output_text)

messages → input
system message → instructions or a developer item
choices[0].message.content → response.output_text
for tools and multimodal output, inspect response.output by item type.

GPT-4.5 Preview - Deprecated

~~GPT-4.5~~ is OpenAI's an advanced preview model, offering state-of-the-art performance across text, code, vision, and reasoning.

Feature	Details
Context window	128,000 input tokens, 16,384 output tokens
Training data	Assumed latest available
Input pricing	$75.00 / 1M tokens
Output pricing	$150.00 / 1M tokens
Strengths	Top-tier reasoning, multimodal understanding, complex task solving
Best for	Cutting-edge applications requiring the highest level of capability, research

GPT-4o

GPT-4o ("omni") is OpenAI's flagship multimodal model, balancing high capability with optimized cost and speed.

Feature	Details
Context window	128,000 input tokens, 16,384 output tokens
Training data	Up to late 2023 (assumed)
Input pricing	$2.50 / 1M tokens
Output pricing	$10.00 / 1M tokens
Strengths	Strong performance across text, code, vision; web search capability
Best for	Complex reasoning, multimodal applications, creative content generation, general tasks

python

response = client.chat.completions.create(
    model="gpt-4o",  # Or specific snapshot like gpt-4o-2024-08-06
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "Explain quantum computing in simple terms."},
    ],
)

Responses API version This version uses `gpt-5.5` because `gpt-4o` may not be enabled for `/v1/responses` in the current AvalAI model data.

Use this version when the selected model supports /v1/responses. messages moves to input, and the final text is read from response.output_text.

python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AVALAI_API_KEY"],
    base_url="https://api.avalai.ir/v1",
)

response = client.responses.create(
    model="gpt-5.5",
    instructions="You are a helpful assistant.",
    input="Explain quantum computing in simple terms.",
)

print(response.output_text)

messages → input
system message → instructions or a developer item
choices[0].message.content → response.output_text
for tools and multimodal output, inspect response.output by item type.

GPT-4o Search Preview

A variant of GPT-4o optimized for tasks involving web search integration.

Feature	Details
Context window	128,000 input tokens, 16,384 output tokens
Training data	Up to late 2023 (assumed), with enhanced search integration
Input pricing	(Check provider details via AvalAI)
Output pricing	(Check provider details via AvalAI)
Max Req/min	50.0
Max Tokens/min	100,000.0
Strengths	Integrates web search results directly into responses
Best for	Questions requiring up-to-date information, research tasks
Model IDs	`gpt-4o-search-preview`, `gpt-4o-search-preview-2025-03-11`

python

response = client.chat.completions.create(
    model="gpt-4o-search-preview",  # Or specific snapshot
    messages=[
        {
            "role": "user",
            "content": "What are the latest developments in fusion energy research?",
        },
    ],
)

Responses API version This version uses `gpt-5.5` because `gpt-4o-search-preview` may not be enabled for `/v1/responses` in the current AvalAI model data.

Use this version when the selected model supports /v1/responses. messages moves to input, and the final text is read from response.output_text.

python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AVALAI_API_KEY"],
    base_url="https://api.avalai.ir/v1",
)

response = client.responses.create(
    model="gpt-5.5",
    instructions="You are a helpful assistant.",
    input="What are the latest developments in fusion energy research?",
)

print(response.output_text)

messages → input
system message → instructions or a developer item
choices[0].message.content → response.output_text
for tools and multimodal output, inspect response.output by item type.

GPT-4o mini

GPT-4o mini offers intelligence comparable to GPT-4 Turbo at a significantly lower price and higher speed, including multimodal capabilities.

Feature	Details
Context window	128,000 input tokens, 16,384 output tokens
Training data	Up to late 2023 (assumed)
Input pricing	$0.15 / 1M tokens
Output pricing	$0.60 / 1M tokens
Strengths	Very fast, highly cost-effective, strong general capabilities, vision, web search
Best for	High-throughput tasks, chatbots, summarization, content moderation, simple analysis

python

response = client.chat.completions.create(
    model="gpt-5.4-mini",
    messages=[
        {"role": "system", "content": "You are a customer support chatbot."},
        {"role": "user", "content": "What are your return policies?"},
    ],
)

Responses API version

Use this version when the selected model supports /v1/responses. messages moves to input, and the final text is read from response.output_text.

python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AVALAI_API_KEY"],
    base_url="https://api.avalai.ir/v1",
)

response = client.responses.create(
    model="gpt-5.4-mini",
    instructions="You are a helpful assistant.",
    input="What are your return policies?",
)

print(response.output_text)

messages → input
system message → instructions or a developer item
choices[0].message.content → response.output_text
for tools and multimodal output, inspect response.output by item type.

GPT-4.1

GPT-4.1 is OpenAI's flagship model for complex tasks, well suited for problem solving across domains.

Feature	Details
Context window	1,047,576 input tokens, 32,768 output tokens
Training data	Up to May 31, 2024
Input pricing	$2.00 / 1M tokens
Output pricing	$8.00 / 1M tokens
Strengths	Strong problem-solving capabilities across domains
Best for	Complex tasks requiring deep understanding and reasoning

python

response = client.chat.completions.create(
    model="gpt-4.1",  # Or specific snapshot like gpt-4.1-2025-04-14
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {
            "role": "user",
            "content": "Analyze the implications of recent advances in quantum computing.",
        },
    ],
)

Responses API version

Use this version when the selected model supports /v1/responses. messages moves to input, and the final text is read from response.output_text.

python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AVALAI_API_KEY"],
    base_url="https://api.avalai.ir/v1",
)

response = client.responses.create(
    model="gpt-4.1",
    instructions="You are a helpful assistant.",
    input="Analyze the implications of recent advances in quantum computing.",
)

print(response.output_text)

messages → input
system message → instructions or a developer item
choices[0].message.content → response.output_text
for tools and multimodal output, inspect response.output by item type.

GPT-4.1-mini

GPT-4.1-mini provides a balance between intelligence, speed, and cost that makes it an attractive model for many use cases.

Feature	Details
Context window	1,047,576 input tokens, 32,768 output tokens
Training data	Up to May 31, 2024
Input pricing	$0.40 / 1M tokens
Output pricing	$1.60 / 1M tokens
Strengths	Balanced performance, good speed and cost efficiency
Best for	General-purpose tasks requiring good performance at moderate cost

python

response = client.chat.completions.create(
    model="gpt-4.1-mini",  # Or specific snapshot like gpt-4.1-mini-2025-04-14
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {
            "role": "user",
            "content": "Provide a summary of the latest climate change research.",
        },
    ],
)

Responses API version

Use this version when the selected model supports /v1/responses. messages moves to input, and the final text is read from response.output_text.

python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AVALAI_API_KEY"],
    base_url="https://api.avalai.ir/v1",
)

response = client.responses.create(
    model="gpt-4.1-mini",
    instructions="You are a helpful assistant.",
    input="Provide a summary of the latest climate change research.",
)

print(response.output_text)

messages → input
system message → instructions or a developer item
choices[0].message.content → response.output_text
for tools and multimodal output, inspect response.output by item type.

GPT-4.1-nano

GPT-4.1-nano is the fastest, most cost-effective GPT-4.1 model, offering good performance for everyday tasks.

Feature	Details
Context window	1,047,576 input tokens, 32,768 output tokens
Training data	Up to May 31, 2024
Input pricing	$0.10 / 1M tokens
Output pricing	$0.40 / 1M tokens
Strengths	Very fast, highly cost-effective
Best for	High-volume applications, simple tasks, content generation

python

response = client.chat.completions.create(
    model="gpt-4.1-nano",  # Or specific snapshot like gpt-4.1-nano-2025-04-14
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {
            "role": "user",
            "content": "Write a short blog post about sustainable living.",
        },
    ],
)

Responses API version

Use this version when the selected model supports /v1/responses. messages moves to input, and the final text is read from response.output_text.

python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AVALAI_API_KEY"],
    base_url="https://api.avalai.ir/v1",
)

response = client.responses.create(
    model="gpt-4.1-nano",
    instructions="You are a helpful assistant.",
    input="Write a short blog post about sustainable living.",
)

print(response.output_text)

messages → input
system message → instructions or a developer item
choices[0].message.content → response.output_text
for tools and multimodal output, inspect response.output by item type.

GPT-4o mini Search Preview

A variant of GPT-4o mini optimized for tasks involving web search integration, offering speed and cost-effectiveness.

Feature	Details
Context window	128,000 input tokens, 16,384 output tokens
Training data	Up to late 2023 (assumed), with enhanced search integration
Input pricing	(Check provider details via AvalAI)
Output pricing	(Check provider details via AvalAI)
Max Req/min	50.0
Max Tokens/min	100,000.0
Strengths	Fast and cost-effective web search integration
Best for	High-throughput queries needing recent information
Model IDs	`gpt-4o-mini-search-preview`, `gpt-4o-mini-search-preview-2025-03-11`

python

response = client.chat.completions.create(
    model="gpt-4o-mini-search-preview",  # Or specific snapshot
    messages=[
        {"role": "user", "content": "Summarize today's top news headlines."},
    ],
)

Responses API version This version uses `gpt-5.5` because `gpt-4o-mini-search-preview` may not be enabled for `/v1/responses` in the current AvalAI model data.

Use this version when the selected model supports /v1/responses. messages moves to input, and the final text is read from response.output_text.

python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AVALAI_API_KEY"],
    base_url="https://api.avalai.ir/v1",
)

response = client.responses.create(
    model="gpt-5.5",
    instructions="You are a helpful assistant.",
    input="Summarize today",
)

print(response.output_text)

messages → input
system message → instructions or a developer item
choices[0].message.content → response.output_text
for tools and multimodal output, inspect response.output by item type.

O1

O1 is a high-intelligence reasoning model focused on complex problem-solving and agency tasks.

Feature	Details
Context window	200,000 input tokens, 100,000 output tokens
Training data	Assumed latest available
Input pricing	$15.00 / 1M tokens
Output pricing	$60.00 / 1M tokens
Strengths	Advanced reasoning, step-by-step task execution, high intelligence
Best for	Complex reasoning, agentic workflows, problem-solving requiring deep analysis

python

response = client.chat.completions.create(
    model="o1",  # Or specific snapshot like o1-2024-12-17
    messages=[
        {"role": "system", "content": "You are a logical reasoning expert."},
        {
            "role": "user",
            "content": "Analyze the logical structure of this argument and identify any fallacies.",
        },
    ],
)

Responses API version This version uses `gpt-5.5` because `o1` may not be enabled for `/v1/responses` in the current AvalAI model data.

Use this version when the selected model supports /v1/responses. messages moves to input, and the final text is read from response.output_text.

python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AVALAI_API_KEY"],
    base_url="https://api.avalai.ir/v1",
)

response = client.responses.create(
    model="gpt-5.5",
    instructions="You are a helpful assistant.",
    input="Analyze the logical structure of this argument and identify any fallacies.",
)

print(response.output_text)

messages → input
system message → instructions or a developer item
choices[0].message.content → response.output_text
for tools and multimodal output, inspect response.output by item type.

O1 mini

O1 mini provides strong reasoning capabilities optimized for speed and cost, suitable for many reasoning-intensive tasks.

Feature	Details
Context window	128,000 input tokens, 65,536 output tokens
Training data	Assumed latest available
Input pricing	$1.10 / 1M tokens
Output pricing	$4.40 / 1M tokens
Strengths	Fast, cost-effective reasoning, good for step-by-step tasks
Best for	Mid-complexity reasoning, structured data extraction, task decomposition

python

response = client.chat.completions.create(
    model="o1-mini",  # Or specific snapshot like o1-mini-2024-09-12
    messages=[
        {"role": "system", "content": "You assist with planning complex tasks."},
        {
            "role": "user",
            "content": "Break down the steps needed to organize a large conference.",
        },
    ],
)

Responses API version This version uses `gpt-5.5` because `o1-mini` may not be enabled for `/v1/responses` in the current AvalAI model data.

Use this version when the selected model supports /v1/responses. messages moves to input, and the final text is read from response.output_text.

python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AVALAI_API_KEY"],
    base_url="https://api.avalai.ir/v1",
)

response = client.responses.create(
    model="gpt-5.5",
    instructions="You are a helpful assistant.",
    input="Break down the steps needed to organize a large conference.",
)

print(response.output_text)

messages → input
system message → instructions or a developer item
choices[0].message.content → response.output_text
for tools and multimodal output, inspect response.output by item type.

GPT-4 Turbo

GPT-4 Turbo was a previous flagship model, now largely superseded by GPT-4o but still available. It offers strong capabilities with a large context window.

Feature	Details
Context window	128,000 tokens input, 4,096 output tokens
Training data	Up to December 2023
Input pricing	$10.00 / 1M tokens
Output pricing	$30.00 / 1M tokens
Strengths	Strong reasoning, instruction following, vision capabilities
Best for	Legacy applications, tasks requiring GPT-4 level quality where GPT-4o is not suitable

python

response = client.chat.completions.create(
    model="gpt-4.1",  # Or specific snapshot like gpt-4.1-2025-04-14
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {
            "role": "user",
            "content": "Write a detailed analysis of renewable energy trends.",
        },
    ],
)

Responses API version

Use this version when the selected model supports /v1/responses. messages moves to input, and the final text is read from response.output_text.

python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AVALAI_API_KEY"],
    base_url="https://api.avalai.ir/v1",
)

response = client.responses.create(
    model="gpt-4.1",
    instructions="You are a helpful assistant.",
    input="Write a detailed analysis of renewable energy trends.",
)

print(response.output_text)

messages → input
system message → instructions or a developer item
choices[0].message.content → response.output_text
for tools and multimodal output, inspect response.output by item type.

GPT-3.5 Turbo

GPT-3.5 Turbo is a highly cost-effective model optimized for speed, suitable for a wide range of simpler tasks.

Feature	Details
Context window	16,385 input tokens, 4,096 output tokens
Training data	Up to September 2021
Input pricing	$1.50 / 1M tokens
Output pricing	$2.00 / 1M tokens
Strengths	Very fast, extremely cost-effective, good for basic tasks
Best for	Simple chatbots, content generation, summarization, classification

python

response = client.chat.completions.create(
    model="gpt-3.5-turbo",  # Or specific snapshot like gpt-3.5-turbo-1106
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "Summarize the key features of machine learning."},
    ],
)

Responses API version This version uses `gpt-5.5` because `gpt-3.5-turbo` may not be enabled for `/v1/responses` in the current AvalAI model data.

Use this version when the selected model supports /v1/responses. messages moves to input, and the final text is read from response.output_text.

python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AVALAI_API_KEY"],
    base_url="https://api.avalai.ir/v1",
)

response = client.responses.create(
    model="gpt-5.5",
    instructions="You are a helpful assistant.",
    input="Summarize the key features of machine learning.",
)

print(response.output_text)

messages → input
system message → instructions or a developer item
choices[0].message.content → response.output_text
for tools and multimodal output, inspect response.output by item type.

Open-Source Models

GPT-OSS-120B

OpenAI's most powerful open-weight model, designed for high-end reasoning tasks and fitting within a single H100 GPU (117B parameters with 5.1B active parameters).

Feature	Details
Context window	131,072 input tokens, 131,072 output tokens
Training data	Up to May 31, 2024
Input pricing	$0.30 / 1M tokens (Azure AI)
Output pricing	$2.50 / 1M tokens (Azure AI)
Provider	Azure AI
License	Apache 2.0
Strengths	Configurable reasoning effort, full chain-of-thought, fine-tunable, agentic capabilities
Best for	Complex reasoning tasks, local deployment, specialized use cases requiring customization

python

response = client.chat.completions.create(
    model="gpt-oss-120b",
    messages=[
        {"role": "system", "content": "You are an expert problem solver."},
        {
            "role": "user",
            "content": "Analyze this complex algorithmic problem and provide a step-by-step solution.",
        },
    ],
)

Responses API version This version uses `gpt-5.5` because `gpt-oss-120b` may not be enabled for `/v1/responses` in the current AvalAI model data.

Use this version when the selected model supports /v1/responses. messages moves to input, and the final text is read from response.output_text.

python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AVALAI_API_KEY"],
    base_url="https://api.avalai.ir/v1",
)

response = client.responses.create(
    model="gpt-5.5",
    instructions="You are a helpful assistant.",
    input="Analyze this complex algorithmic problem and provide a step-by-step solution.",
)

print(response.output_text)

messages → input
system message → instructions or a developer item
choices[0].message.content → response.output_text
for tools and multimodal output, inspect response.output by item type.

GPT-OSS-120B (AWS Bedrock)

The same powerful 120B parameter model available through AWS Bedrock with competitive pricing.

Feature	Details
Context window	131,072 input tokens, 131,072 output tokens
Training data	Up to May 31, 2024
Input pricing	$0.15 / 1M tokens (AWS Bedrock)
Output pricing	$0.60 / 1M tokens (AWS Bedrock)
Provider	AWS Bedrock
License	Apache 2.0
Strengths	More cost-effective pricing, configurable reasoning effort, full chain-of-thought
Best for	Cost-sensitive applications requiring high-end reasoning capabilities

python

response = client.chat.completions.create(
    model="openai.gpt-oss-120b-1:0",
    messages=[
        {"role": "system", "content": "You are a research assistant."},
        {
            "role": "user",
            "content": "Conduct a thorough analysis of the given research paper and identify key insights.",
        },
    ],
)

Responses API version This version uses `gpt-5.5` because `openai.gpt-oss-120b-1:0` may not be enabled for `/v1/responses` in the current AvalAI model data.

Use this version when the selected model supports /v1/responses. messages moves to input, and the final text is read from response.output_text.

python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AVALAI_API_KEY"],
    base_url="https://api.avalai.ir/v1",
)

response = client.responses.create(
    model="gpt-5.5",
    instructions="You are a helpful assistant.",
    input="Conduct a thorough analysis of the given research paper and identify key insights.",
)

print(response.output_text)

messages → input
system message → instructions or a developer item
choices[0].message.content → response.output_text
for tools and multimodal output, inspect response.output by item type.

GPT-OSS-20B

A medium-sized open-weight model optimized for low latency and local deployment (21B parameters with 3.6B active parameters).

Feature	Details
Context window	131,072 input tokens, 131,072 output tokens
Training data	Up to May 31, 2024
Input pricing	$0.07 / 1M tokens (AWS Bedrock)
Output pricing	$0.30 / 1M tokens (AWS Bedrock)
Provider	AWS Bedrock
License	Apache 2.0
Strengths	Fast inference, cost-effective, configurable reasoning effort, fine-tunable
Best for	Low latency applications, high-volume processing, local deployment scenarios

python

response = client.chat.completions.create(
    model="openai.gpt-oss-20b-1:0",
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {
            "role": "user",
            "content": "Summarize this document and extract the key action items.",
        },
    ],
)

Responses API version This version uses `gpt-5.5` because `openai.gpt-oss-20b-1:0` may not be enabled for `/v1/responses` in the current AvalAI model data.

Use this version when the selected model supports /v1/responses. messages moves to input, and the final text is read from response.output_text.

python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AVALAI_API_KEY"],
    base_url="https://api.avalai.ir/v1",
)

response = client.responses.create(
    model="gpt-5.5",
    input=[
        {
            "role": "user",
            "content": [
                {
                    "type": "input_text",
                    "text": "Summarize this document and extract the key action items.",
                },
                {"type": "input_file", "file_id": "file_abc123"},
            ],
        }
    ],
)

print(response.output_text)

messages → input
system message → instructions or a developer item
choices[0].message.content → response.output_text
for tools and multimodal output, inspect response.output by item type.

Key Features of Open-Source Models

All OpenAI open-source models share these powerful capabilities:

Permissive Apache 2.0 License: Build freely without copyleft restrictions or patent risk
Configurable Reasoning Effort: Adjust reasoning effort (low, medium, high) based on your needs
Full Chain-of-Thought: Complete access to the model's reasoning process for debugging and trust
Fine-Tunable: Customize models to your specific use case through parameter fine-tuning
Agentic Capabilities: Native support for function calling, web browsing, Python code execution, and structured outputs

Video Generation Models

Sora 2

Sora 2 is OpenAI's video generation model optimized for speed and flexibility, ideal for rapid iteration and content exploration. Accessible through the v1/videos endpoint only.

Feature	Details
Resolution options	720x1280 (portrait), 1280x720 (landscape)
Duration	4-8+ seconds
Pricing	$0.10 per second of generated video
Strengths	Fast generation, cost-effective, good quality for prototyping
Best for	Social media content, rapid iteration, prototypes, concepting
Model ID	`sora-2`
Important	Only works with v1/videos endpoint

python

from openai import OpenAI

client = OpenAI(api_key="your-avalai-api-key", base_url="https://api.avalai.ir/v1")

# Create a video generation job
video = client.videos.create(
    model="sora-2",
    prompt="A calico cat playing a piano on stage, cinematic lighting",
    size="720x1280",
    seconds=4,
)

print(f"Video job created: {video.id}")

# Poll for completion
video_status = client.videos.retrieve(video.id)
while video_status.status in ["queued", "in_progress"]:
    time.sleep(10)
    video_status = client.videos.retrieve(video.id)
    print(f"Status: {video_status.status}, Progress: {video_status.progress}%")

# Download completed video
if video_status.status == "completed":
    content = client.videos.download_content(video.id, variant="video")
    content.write_to_file("output.mp4")

Key Features:

Asynchronous Generation: Jobs are processed asynchronously with status polling or webhook notifications
Image References: Provide input images to guide video generation
Remix Capability: Iterate on completed videos with targeted adjustments
Supporting Assets: Download thumbnails and spritesheets alongside videos

Sora 2 Pro

Sora 2 Pro is OpenAI's premium video generation model designed for production-quality output with enhanced visual fidelity. Accessible through the v1/videos endpoint only.

Feature	Details
Resolution options	720x1280, 1280x720, 1024x1792, 1792x1024
Duration	4-8+ seconds
Pricing	$0.30/second (720x1280, 1280x720), $0.50/second (1024x1792, 1792x1024)
Strengths	High-quality output, cinematic results, stable motion, detailed rendering
Best for	Marketing assets, cinematic footage, production-quality content
Model ID	`sora-2-pro`
Important	Only works with v1/videos endpoint

python

from openai import OpenAI

client = OpenAI(api_key="your-avalai-api-key", base_url="https://api.avalai.ir/v1")

# Create a high-quality video
video = client.videos.create(
    model="sora-2-pro",
    prompt="Wide tracking shot of a teal coupe driving through a desert highway, heat ripples visible, hard sun overhead, cinematic quality",
    size="1280x720",
    seconds=8,
)

# Use webhooks or polling to monitor completion
# Once completed, download the video
content = client.videos.download_content(video.id)
content.write_to_file("professional_video.mp4")

Advanced Features:

Higher Resolution Support: Includes 1024x1792 and 1792x1024 for ultra-high-quality output
Enhanced Motion Stability: Improved camera movement and object tracking
Cinematic Quality: Superior lighting, color grading, and visual consistency
Longer Generation Time: Takes more time but produces polished results

For detailed video generation workflows, see the Video Generation Guide.

Image Generation Models

GPT-Image-2

GPT-Image-2 is OpenAI's next-generation image generation model (ChatGPT Images 2.0), delivering greater precision and control, stronger multilingual text rendering, stylistic sophistication and realism, and support for flexible aspect ratios. It leverages thinking mode to research, transform inputs, and produce cohesive, end-to-end visual assets.

Feature	Details
Model ID	`gpt-image-2`
Aliases	`gpt-image-2-2026-04-21`
Supported Endpoints	`v1/images/generations`, `v1/images/edits`
Input	Text, Image
Output	Image, Text
Text Input pricing	$5.00 / 1M tokens
Text Cached Input pricing	$1.25 / 1M tokens
Text Output pricing	$10.00 / 1M tokens
Image Input pricing	$8.00 / 1M tokens
Image Cached Input pricing	$2.00 / 1M tokens
Image Output pricing	$30.00 / 1M tokens
Strengths	Greater precision & control, multilingual text rendering, stylistic sophistication, flexible aspect ratios, thinking mode
Best for	Professional design, editorial posters, global typography, stylized illustrations, complex visual storytelling

Key Capabilities:

Greater Precision & Control — significantly improved accuracy and adherence to prompt details
Multilingual Text Rendering — strong performance across Japanese, Arabic, Korean, Devanagari, Cyrillic, Bengali, Greek, Chinese, and Latin scripts
Stylistic Sophistication & Realism — higher fidelity across photography, illustration, manga, pixel art, and other styles
Flexible Aspect Ratios — Horizontal, Square, and Vertical formats for banners, posts, and mobile screens
Thinking Mode — uses reasoning to research, transform inputs, and produce cohesive, end-to-end visual assets
Image Editing — supports both v1/images/generations and v1/images/edits endpoints

python

response = client.images.generate(
    model="gpt-image-2",
    prompt='A modernist editorial poster titled "Typography" celebrating global scripts including Japanese, Arabic, Korean, and Latin letterforms, in red, blue, and black tones.',
    size="1024x1024",
    n=1,
)

GPT Image 2 Cost Calculation for Image Edits

When editing images with gpt-image-2 through v1/images/edits, billing is purely token-based. There is no flat per-edit fee: you pay for the prompt text tokens, the image input tokens for every reference image you upload, and the image output tokens the model returns. The quality and size you request drive the output image tokens, which usually dominate the cost of an edit.

Approximate Per-Image Costs by Quality and Resolution

These are calculator estimates from OpenAI's image generation cost tool, not fixed flat rates. Actual costs vary based on prompt complexity, image size, and whether reference images are included.

Quality	1024x1024 (square)	1024x1536 (portrait)	1536x1024 (landscape)
Low	~$0.008	~$0.012	~$0.012
Medium	~$0.032	~$0.048	~$0.048
High	~$0.125	~$0.187	~$0.187

These estimates cover output image tokens only, billed at the GPT Image 2 image output rate of $30.00 / 1M tokens. Edits also add prompt text tokens ($5.00 / 1M) and image input tokens for every reference image you upload ($8.00 / 1M, or $2.00 / 1M when cached). Always use OpenAI's official image generation calculator with your specific prompt and settings for exact cost estimates.

What Each Quality Setting Actually Produces

Quality settings are not just a dial between worse and better. Each tier suits specific use cases and choosing the right one has a direct impact on monthly costs:

Low quality: Faster generation at the lowest cost. Works for drafts, small-format digital assets, and automated pipelines where cost control matters more than fine detail.
Medium quality: Appropriate for most marketing and content production: social media imagery, editorial graphics, product mockups, and campaign assets. Sufficient for professional publication in the majority of contexts.
High quality: Designed for final production assets where pixel-level accuracy matters: hero product photography, high-resolution print materials, packaging designs, and detailed UI mockups.

For detailed edit cost-control patterns, see Generate Images with GPT Image.

GPT-Image-1

GPT-Image-1 is OpenAI's latest image generation model, designed for high-quality, photorealistic image creation with exceptional prompt following.

Feature	Details
Sizes	1024x1024, 1024x1792, 1792x1024, 1792x1792
Quality options	Low, Medium, High
Pricing	$0.011 / image (Low), $0.042 / image (Medium), $0.167 / image (High)
Strengths	Photorealistic images, strong prompt adherence, consistent style
Best for	Creating high-quality images from detailed text descriptions

python

response = client.images.generate(
    model="gpt-image-1",
    prompt="A photorealistic image of a futuristic city with flying cars and tall skyscrapers covered in vegetation, sunset lighting, highly detailed",
    size="1024x1024",
    quality="medium",
    n=1,
    response_format="url",  # or b64_json
)

GPT-Image-1.5

GPT-Image-1.5 is OpenAI's latest image generation model, delivering major improvements in realism, accuracy, and editability compared to GPT-Image-1. It supports high-fidelity photorealism, text rendering, style transfer, and complex structured visuals like infographics.

Feature	Details
Supported Endpoints	`v1/images/generations`, `v1/images/edits`
Input	Text, Image
Output	Image, Text
Quality options	Low, Medium, High
Text Input pricing	$5.00 / 1M tokens
Text Cached Input pricing	$2.00 / 1M tokens
Text Output pricing	$32.00 / 1M tokens
Image Input pricing	$8.00 / 1M tokens
Image Output pricing	$32.00 / 1M tokens
Strengths	State-of-the-art photorealism, text rendering in images, style transfer, infographics
Best for	Professional design, marketing creatives, infographics, product mockups, identity-preserved editing

Key Capabilities:

High-fidelity photorealism with natural lighting, accurate materials, and rich color rendering
Reliable text rendering with crisp lettering, consistent layout, and strong contrast inside images
Complex structured visuals including infographics, diagrams, and multi-panel compositions
Precise style control and style transfer with minimal prompting
Robust facial and identity preservation for edits, character consistency, and multi-step workflows

python

# Image Generation
response = client.images.generate(
    model="gpt-image-1.5",
    prompt="A detailed infographic explaining how a coffee machine works, with labeled components and flow diagram",
    size="1024x1024",
    quality="high",
    n=1,
    response_format="b64_json",
)

# Image Editing
with open("input_image.png", "rb") as image_file:
    response = client.images.edit(
        model="gpt-image-1.5",
        image=image_file,
        prompt="Make it look like a winter evening with snowfall",
        size="1024x1024",
    )

GPT-Image-1-Mini

GPT-Image-1-Mini is a cost-efficient version of GPT-Image-1, offering faster and more affordable image generation while maintaining high-quality, photorealistic output. Available for Tier 3, 4, and 5 users.

Feature	Details
Sizes	1024x1024, 1024x1536, 1536x1024
Quality options	Low, Medium, High
Pricing	$0.011 / image (Low 1024x1024), $0.042 / image (Medium 1024x1024), $0.036 / image (High 1024x1024)
Token pricing	Input: $5.00/1M, Cached: $1.25/1M, Output: $40.00/1M
Strengths	Cost-effective, fast generation, photorealistic images, strong prompt adherence
Best for	High-volume image generation, prototyping, cost-sensitive applications
Availability	Tier 3, 4, 5 users

python

response = client.images.generate(
    model="gpt-image-1-mini",
    prompt="A photorealistic image of a mountain landscape with a lake reflecting the sunset",
    size="1024x1024",
    quality="high",
    n=1,
    response_format="url",  # or b64_json
)

Embedding Models

text-embedding-3-large

OpenAI's most capable embedding model for tasks requiring nuanced understanding.

Feature	Details
Dimensions	3072 (default) or configurable
Max input tokens	8191
Pricing	$0.13 / 1M tokens
Strengths	Highest quality embeddings for semantic search, clustering, classification
Best for	Applications demanding maximum accuracy in text similarity and understanding

python

response = client.embeddings.create(
    model="text-embedding-3-large",
    input="The food was delicious and the service was excellent.",
)

text-embedding-3-small

A highly efficient embedding model offering a great balance of performance and cost.

Feature	Details
Dimensions	1536 (default) or configurable
Max input tokens	8191

GPT-4o Transcribe & GPT-4o mini Transcribe

Specialized variants of GPT-4o and GPT-4o mini fine-tuned for high-accuracy speech-to-text transcription tasks. These offer potentially improved accuracy or efficiency over the general Whisper model for transcription.

Feature	Details (Applies to both unless noted)
Input formats	mp3, mp4, mpeg, mpga, m4a, wav, webm
Max file size	(Check provider details - likely similar to Whisper)
Pricing	(Check provider details via AvalAI)
Max Req/min	500.0
Max Tokens/min	200,000.0
Strengths	Optimized for transcription accuracy and/or efficiency
Best for	High-quality transcription needs, potentially lower latency/cost than Whisper
Model IDs	`gpt-4o-transcribe`, `gpt-4o-mini-transcribe`

python

# Example using GPT-4o Transcribe
with open("meeting_audio.mp3", "rb") as audio_file:
    transcription = client.audio.transcriptions.create(
        model="gpt-4o-transcribe",  # Or gpt-4o-mini-transcribe
        file=audio_file,
        # Additional parameters like language might be supported
    )

# Example using GPT-4o Mini Transcribe
with open("short_clip.wav", "rb") as audio_file:
    transcription = client.audio.transcriptions.create(
        model="gpt-4o-mini-transcribe",
        file=audio_file,
    )

GPT-4o Transcribe Diarize

Enhanced version of gpt-4o-transcribe with advanced speaker diarization capabilities, enabling accurate identification and separation of multiple speakers in audio recordings.

Feature	Details
Input formats	mp3, mp4, mpeg, mpga, m4a, wav, webm
Max file size	(Check provider details - likely similar to Whisper)
Text input pricing	$2.50 / 1M tokens
Audio input pricing	$6.00 / 1M tokens
Cached input pricing	$1.50 / 1M tokens
Output pricing	$10.00 / 1M tokens
Max Req/min	500.0
Max Tokens/min	200,000.0
Strengths	Speaker diarization, fast transcription (~15s for 10min audio), 100+ languages
Best for	Meeting transcription, interview analysis, multi-speaker content, customer service calls
Model ID	`gpt-4o-transcribe-diarize`

Note

This model provides the same pricing as gpt-4o-transcribe while offering enhanced speaker identification capabilities through the verbose_json response format.

python

# Example using GPT-4o Transcribe Diarize with speaker identification
with open("meeting_audio.mp3", "rb") as audio_file:
    transcription = client.audio.transcriptions.create(
        model="gpt-4o-transcribe-diarize",
        file=audio_file,
        response_format="verbose_json",  # Required for speaker diarization
    )

# Access speaker information from segments
for segment in transcription.segments:
    print(f"{segment.speaker}: {segment.text}")
    # Output example:
    # SPEAKER_1: Hello, welcome to the meeting.
    # SPEAKER_2: Thank you, glad to be here.

Use Cases:

Meeting Analysis: Automatically transcribe and attribute speaker contributions for meeting summaries
Customer Service: Analyze customer-agent interactions with accurate speaker attribution
Content Production: Transcribe interviews and podcasts with speaker labels for post-production
Compliance Monitoring: Track individual participation and speaking time in recorded conversations

python

response = client.embeddings.create(
    model="text-embedding-3-small",
    input="The food was delicious and the service was excellent.",
)

text-embedding-ada-002

Legacy embedding model, primarily for backward compatibility. text-embedding-3-small is recommended for new applications.

Feature	Details
Dimensions	1536
Max input tokens	8191
Pricing	$0.10 / 1M tokens
Strengths	Consistent performance for legacy systems
Best for	Maintaining compatibility with systems built on `ada-002`

python

response = client.embeddings.create(
    model="text-embedding-ada-002",
    input="The food was delicious and the service was excellent.",
)

Audio Models

GPT-Audio-1.5

GPT-Audio-1.5 is OpenAI's best voice model for audio in, audio out with Chat Completions. It's optimized for real-time interaction with low latency and natural-sounding speech, making it ideal for voice agents and conversational AI applications.

Feature	Details
Context window	128,000 tokens
Max output tokens	16,384 tokens
Knowledge cutoff	September 30, 2024
Text input pricing	$2.50 / 1M tokens
Cached input pricing	$1.25 / 1M tokens
Text output pricing	$10.00 / 1M tokens
Audio input pricing	$32.00 / 1M tokens
Audio output pricing	$64.00 / 1M tokens
Input modalities	Text, Audio
Output modalities	Text, Audio
Strengths	Low latency, natural speech, improved instruction following, function calling
Best for	Voice agents, real-time translators, interactive assistants
Model IDs	`gpt-audio-1.5`

Key Features:

Low Latency & High Speed: Optimized for real-time interaction
Natural Sounding Speech: Smoother, more conversational audio with enhanced pacing and prosody
Improved Instruction Following: Better alignment with user and developer instructions
Function Calling: Support for interactive, tool-driven applications
Multiple Voice Options: Clear and consistent audio across different voices

python

response = client.chat.completions.create(
    model="gpt-audio-1.5",
    messages=[
        {
            "role": "user",
            "content": "Tell me a short story about a robot learning to paint.",
        }
    ],
    modalities=["text", "audio"],
    audio={"format": "mp3", "voice": "nova"},
)

# Access audio data
audio_data = response.choices[0].message.audio.data
transcript = response.choices[0].message.audio.transcript

Responses API version

Use gpt-audio for a Responses-compatible text-first migration path, because gpt-audio-1.5 is currently documented here as a Chat Completions audio model.

python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AVALAI_API_KEY"],
    base_url="https://api.avalai.ir/v1",
)

response = client.responses.create(
    model="gpt-audio",
    input="Tell me a short story about a robot learning to paint.",
)

print(response.output_text)

messages → input
choices[0].message.content → response.output_text
keep Chat Completions for audio output when the target model exposes only /v1/chat/completions

GPT Audio

GPT Audio is OpenAI's generally available audio model for conversational AI applications. It supports both audio and text inputs and outputs through the Chat Completions API.

Feature	Details
Context window	128,000 tokens
Max output tokens	16,384 tokens
Training data	Up to October 1, 2023
Text input pricing	$2.50 / 1M tokens
Cached input pricing	$1.25 / 1M tokens
Text output pricing	$10.00 / 1M tokens
Audio input pricing	$32.00 / 1M tokens
Audio output pricing	$64.00 / 1M tokens
Strengths	Native audio processing, multimodal conversations, function calling
Best for	Voice assistants, conversational apps, audio-based interactions
Model IDs	`gpt-audio`, `gpt-audio-2025-08-28`

python

response = client.chat.completions.create(
    model="gpt-audio",
    messages=[{"role": "user", "content": "Tell me about artificial intelligence."}],
    modalities=["text", "audio"],
    audio={"format": "mp3", "voice": "nova"},
)

# Access audio data
audio_data = response.choices[0].message.audio.data
transcript = response.choices[0].message.audio.transcript

Responses API version

Use this version when you want the same prompt through /v1/responses.

python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AVALAI_API_KEY"],
    base_url="https://api.avalai.ir/v1",
)

response = client.responses.create(
    model="gpt-audio",
    input="Tell me about artificial intelligence.",
)

print(response.output_text)

messages → input
choices[0].message.content → response.output_text
inspect response.output for non-text response items

GPT Audio Mini

GPT Audio Mini is a cost-efficient version of GPT Audio, ideal for high-volume audio processing applications.

Feature	Details
Context window	128,000 tokens
Max output tokens	16,384 tokens
Training data	Up to October 1, 2023
Text input pricing	$0.60 / 1M tokens
Cached input pricing	$0.30 / 1M tokens
Text output pricing	$2.40 / 1M tokens
Audio input pricing	$10.00 / 1M tokens
Audio output pricing	$20.00 / 1M tokens
Strengths	Cost-effective audio processing, multimodal, function calling
Best for	High-volume audio applications, cost-sensitive voice interactions
Model IDs	`gpt-audio-mini`, `gpt-audio-mini-2025-10-06`

python

response = client.chat.completions.create(
    model="gpt-audio-mini",
    messages=[{"role": "user", "content": "What is machine learning?"}],
    modalities=["text", "audio"],
    audio={"format": "mp3", "voice": "alloy"},
)

Responses API version

Use this version when you want the same prompt through /v1/responses.

python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AVALAI_API_KEY"],
    base_url="https://api.avalai.ir/v1",
)

response = client.responses.create(
    model="gpt-audio-mini",
    input="What is machine learning?",
)

print(response.output_text)

messages → input
choices[0].message.content → response.output_text
inspect response.output for non-text response items

Legacy Audio Models:

For backwards compatibility, the following preview models remain available:

gpt-4o-audio-preview
gpt-4o-mini-audio-preview

Audio Format Support:

Output formats: mp3, wav, pcm16, opus, aac, flac
Voice options: alloy, echo, fable, onyx, nova, shimmer

Whisper

Whisper is a versatile speech-to-text model for transcription and translation.

Feature	Details
Input formats	mp3, mp4, mpeg, mpga, m4a, wav, webm
Max file size	25 MB
Pricing	$0.006 / minute (rounded to nearest second)
Strengths	Accurate transcription and translation across many languages
Best for	Transcribing meetings, interviews, audio notes; translating spoken content

python

with open("audio.mp3", "rb") as audio_file:
    transcription = client.audio.transcriptions.create(
        model="whisper-1", file=audio_file
    )

GPT-4o mini TTS

GPT-4o mini TTS is a text-to-speech model powered by GPT-4o mini, providing natural-sounding speech synthesis.

Feature	Details
Max input tokens	2,000
Input pricing	$0.60 / 1M tokens
Output pricing	$12.00 / 1M tokens (audio)
Strengths	Fast, high-quality speech synthesis
Best for	Applications requiring natural-sounding voice output

python

response = client.audio.speech.create(
    model="gpt-4o-mini-tts",
    voice="nova",  # Assuming it supports the same voices as standard TTS
    input="This is GPT-4o mini TTS converting text to natural-sounding speech.",
)

TTS (Text-to-Speech)

TTS models convert text into natural-sounding human speech.

Feature	Details
Models	`tts-1` (Real-time optimized), `tts-1-hd` (Quality optimized)
Voices	alloy, echo, fable, onyx, nova, shimmer
Pricing	$15.00 / 1M characters (TTS), $30.00 / 1M characters (TTS HD)
Strengths	High-quality, natural-sounding speech synthesis
Best for	Voice generation for applications, accessibility tools, content creation

python

response = client.audio.speech.create(
    model="tts-1",
    voice="nova",
    input="Hello world! This is a text-to-speech model speaking.",
)

Web Search Capabilities

OpenAI models support advanced web search capabilities that allow them to access up-to-date information from the internet and provide answers with sourced citations. Web search is available through the Responses API using the web_search tool.

Types of Web Search

There are three main types of web search available with OpenAI models:

Non-reasoning web search: The model sends the user's query to the web search tool, which returns the response based on top results. There's no internal planning and the model simply passes along the search tool's responses. This method is fast and ideal for quick lookups.
Agentic search with reasoning models: The model actively manages the search process. It can perform web searches as part of its chain of thought, analyze results, and decide whether to keep searching. This flexibility makes agentic search well suited to complex workflows, but searches take longer than quick lookups.
Deep research: A specialized, agent-driven method for in-depth, extended investigations by reasoning models. The model conducts web searches as part of its chain of thought, often tapping into hundreds of sources. Deep research can run for several minutes and is best used with background mode.

Web Search Tool Configuration

Basic Web Search

python

import requests

url = "https://api.avalai.ir/v1/responses"
headers = {
    "Content-Type": "application/json",
    "Authorization": "Bearer $AVALAI_API_KEY",
}

data = {
    "model": "gpt-5",
    "tools": [{"type": "web_search"}],
    "input": "What was a positive news story from today?",
}

response = requests.post(url, headers=headers, json=data)

Domain Filtering

Domain filtering lets you limit results to a specific set of domains. You can set an allow-list of up to 20 URLs:

python

data = {
    "model": "gpt-5",
    "tools": [
        {
            "type": "web_search",
            "filters": {
                "allowed_domains": [
                    "pubmed.ncbi.nlm.nih.gov",
                    "clinicaltrials.gov",
                    "www.who.int",
                    "www.cdc.gov",
                    "www.fda.gov",
                ]
            },
        }
    ],
    "input": "Please perform a web search on how semaglutide is used in the treatment of diabetes.",
}

User Location

To refine search results based on geography, you can specify an approximate user location:

python

data = {
    "model": "o4-mini",
    "tools": [
        {
            "type": "web_search",
            "user_location": {
                "type": "approximate",
                "country": "GB",
                "city": "London",
                "region": "London",
                "timezone": "Europe/London",
            },
        }
    ],
    "input": "What are the best restaurants around Granary Square?",
}

Sources Field

To view all URLs retrieved during a web search, use the sources field:

python

data = {
    "model": "gpt-5",
    "reasoning": {"effort": "low"},
    "tools": [
        {
            "type": "web_search",
            "filters": {
                "allowed_domains": [
                    "pubmed.ncbi.nlm.nih.gov",
                    "clinicaltrials.gov",
                    "www.who.int",
                ]
            },
        }
    ],
    "tool_choice": "auto",
    "include": ["web_search_call.action.sources"],
    "input": "Please perform a web search on how semaglutide is used in the treatment of diabetes.",
}

Response Format and Citations

Model responses that use the web search tool include two parts:

A web_search_call output item with the ID of the search call, along with the action taken in web_search_call.action. The action is one of:
- search: represents a web search with query and domains searched
- open_page: represents a page being opened (supported in reasoning models)
- find_in_page: represents searching within a page (supported in reasoning models)
A message output item containing:
- The text result in message.content[0].text
- Annotations message.content[0].annotations for the cited URLs

Example Response Structure

json

[
  {
    "type": "web_search_call",
    "id": "ws_67c9fa0502748190b7dd390736892e100be649c1a5ff9609",
    "status": "completed"
  },
  {
    "id": "msg_67c9fa077e288190af08fdffda2e34f20be649c1a5ff9609",
    "type": "message",
    "status": "completed",
    "role": "assistant",
    "content": [
      {
        "type": "output_text",
        "text": "On March 6, 2025, several news...",
        "annotations": [
          {
            "type": "url_citation",
            "start_index": 2606,
            "end_index": 2758,
            "url": "https://...",

            "title": "Title..."
          }
        ]
      }
    ]
  }
]

API Compatibility

Web search is available in the Responses API as:

web_search: The generally available version of the tool
web_search_preview: The earlier tool version

To use web search in the Chat Completions API, use the specialized web search models:

gpt-4o-search-preview
gpt-4o-mini-search-preview

Limitations

Web search is currently not supported in gpt-5 with minimal reasoning and gpt-4.1-nano
When used as a tool in the Responses API, web search has the same tiered rate limits as the underlying models
Web search is limited to a context window size of 128,000 tokens (even with gpt-4.1 and gpt-4.1-mini models)
User location is not supported for deep research models using web search

Moderation Models

text-moderation-latest / omni-moderation-latest

Free models for identifying potentially harmful or policy-violating content in text. omni-moderation is the newest version.

Feature	Details
Categories	Various categories including hate, self-harm, sexual, violence
Pricing	Free
Strengths	Fast, reliable detection of problematic content
Best for	Content filtering, enforcing usage policies, maintaining safe online environments

python

response = client.moderations.create(
    model="omni-moderation-latest",  # Or text-moderation-latest
    input="Sample text to check for harmful content.",
)

Model Selection Guidelines

Choosing the Right Model

Task Type: Identify the primary task (chat, reasoning, image gen, embedding, audio, moderation).
Complexity & Quality: More complex tasks or those needing highest quality benefit from GPT-5, GPT-4.1, o4-mini or O3.
Speed & Cost: For high volume, low latency, or cost-sensitive tasks, GPT-4.1 mini, GPT-5-mini, or O4 mini are strong choices.
Multimodality: If vision or audio input/output is needed, select models like GPT-5-chat, GPT-5 mini, or specialized audio models.
Context Length: Consider the amount of input information the model needs to process (most modern models support 128k+ tokens).

Performance Comparison (General Guide)

Task	Recommended OpenAI Models	Alternative Models (Other Providers)
Complex Reasoning / Research	GPT-5, O3, GPT-4.1	Claude 4.1 Opus, Gemini 2.5 Pro
General Chat / Content Gen	GPT-5-chat, GPT-5 mini	Claude 4/3.7 Sonnet, Gemini 2.5 Flash
Code Generation / Analysis	GPT-5 Chat, O4 mini	Claude 4/3.7 Sonnet, Gemini 2.5 Pro
High Volume / Low Latency	GPT-4o mini, GPT-3.5 Turbo	Claude 3.5 Haiku, Gemini 1.5/2.0 Flash
Vision / Multimodal Tasks	GPT-5 Chat, O3, GPT-5 mini	Claude 3 series, Gemini 2.5 models
Image Generation	DALL·E 3	Stable Diffusion 3, Midjourney
Embeddings (Balanced)	text-embedding-3-small	Cohere Embed v3
Embeddings (Highest Quality)	text-embedding-3-large	N/A
Audio Transcription	Whisper	Gemini models, Claude 4 Sonnet

Using OpenAI Models via AvalAI

All OpenAI models are accessible through the standard AvalAI API endpoints using OpenAI-compatible client libraries:

python

from openai import OpenAI

client = OpenAI(
    api_key="your-avalai-api-key",  # Replace with your actual API key
    base_url="https://api.avalai.ir/v1",  # AvalAI API endpoint
)

# Example: Using GPT-4o mini
response = client.chat.completions.create(
    model="gpt-5.4-mini", messages=[{"role": "user", "content": "Hello!"}]
)

# Example: Generating an image with gpt-image-2
image_response = client.images.generate(
    model="gpt-image-2", prompt="A cute cat wearing a wizard hat", n=1, size="1024x1024"
)

Responses API version

Use Responses for text, reasoning, and tool-capable OpenAI models while keeping Images API calls for direct image generation.

python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["AVALAI_API_KEY"],
    base_url="https://api.avalai.ir/v1",
)

response = client.responses.create(
    model="gpt-5.5",
    input="Hello!",
)

print(response.output_text)

chat.completions.create(...) → responses.create(...)
messages → input
choices[0].message.content → response.output_text

Model Versioning

OpenAI frequently updates models. AvalAI provides access using both general aliases (e.g., gpt-4o) and specific version snapshots (e.g., gpt-4o-2024-08-06). Using snapshots guarantees stability for your application.

Latest Aliases: gpt-5, gpt-5-mini, gpt-5-nano, gpt-5-chat-latest, o4-mini, o1-pro, o3, o3-mini, gpt-4.1, gpt-4.1-mini, gpt-4.1-nano, gpt-4o, gpt-4o-mini, o1, o1-mini, dall-e-3, text-embedding-3-large, text-embedding-3-small, whisper-1, tts-1, omni-moderation-latest
Specific Snapshots: e.g., gpt-5-2025-08-07, gpt-5-mini-2025-08-07, gpt-5-nano-2025-08-07, o1-pro-2025-03-19, o3-2025-04-16, gpt-4.1-2025-04-14, gpt-4o-2024-08-06, gpt-5.4-mini-2026-03-17, o1-2024-12-17

Check the Model Details page for the full list of available snapshots and aliases.

OpenAI Models ​

Chat, Reasoning, and Multimodal Models ​

GPT-5.6 Family ​

GPT-5.5 ​

GPT-5.4 Pro ​

GPT-5.4 ​

GPT-5.4 Mini ​

GPT-5.4 Nano ​

GPT-5.3 Chat ​

GPT-5.2 Pro ​

GPT-5.2 ​

GPT-5.3-Codex ​

GPT-5.2-Codex ​

GPT-5 Pro ​

GPT-5.1 ​

GPT-5.1 Chat ​

GPT-5 ​

GPT-5 Mini ​

GPT-5 Nano ​

GPT-5 Chat ​

GPT-5 Codex ​

GPT-5.1 Codex ​

GPT-5.1 Codex Mini ​

GPT-5.1-Codex-Max ​

o3-pro ​

o3-deep-research ​

Web Search with Domain Filtering ​

o3 ​

o4-mini-deep-research ​

Web Search with User Location ​

o4-mini ​

GPT-4.5 Preview - Deprecated ​

GPT-4o ​

GPT-4o Search Preview ​

GPT-4o mini ​

GPT-4.1 ​

GPT-4.1-mini ​

GPT-4.1-nano ​

GPT-4o mini Search Preview ​

O1 ​

O1 mini ​

GPT-4 Turbo ​

GPT-3.5 Turbo ​

Open-Source Models ​

GPT-OSS-120B ​

GPT-OSS-120B (AWS Bedrock) ​

GPT-OSS-20B ​

Key Features of Open-Source Models ​

Video Generation Models ​

Sora 2 ​

Sora 2 Pro ​

Image Generation Models ​

GPT-Image-2 ​

GPT Image 2 Cost Calculation for Image Edits ​

GPT-Image-1 ​

GPT-Image-1.5 ​

GPT-Image-1-Mini ​

Embedding Models ​

text-embedding-3-large ​

text-embedding-3-small ​

GPT-4o Transcribe & GPT-4o mini Transcribe ​

GPT-4o Transcribe Diarize ​

text-embedding-ada-002 ​

Audio Models ​

GPT-Audio-1.5 ​

GPT Audio ​

GPT Audio Mini ​

Whisper ​

GPT-4o mini TTS ​

TTS (Text-to-Speech) ​

Web Search Capabilities ​

Types of Web Search ​

Web Search Tool Configuration ​

Basic Web Search ​

Domain Filtering ​

User Location ​

Sources Field ​

Response Format and Citations ​

Example Response Structure ​

API Compatibility ​

OpenAI Models

Chat, Reasoning, and Multimodal Models

GPT-5.6 Family

GPT-5.5

GPT-5.4 Pro

GPT-5.4

GPT-5.4 Mini

GPT-5.4 Nano

GPT-5.3 Chat

GPT-5.2 Pro

GPT-5.2

GPT-5.3-Codex

GPT-5.2-Codex

GPT-5 Pro

GPT-5.1

GPT-5.1 Chat

GPT-5

GPT-5 Mini

GPT-5 Nano

GPT-5 Chat

GPT-5 Codex

GPT-5.1 Codex

GPT-5.1 Codex Mini

GPT-5.1-Codex-Max

o3-pro

o3-deep-research

Web Search with Domain Filtering

o3

o4-mini-deep-research

Web Search with User Location

o4-mini

GPT-4.5 Preview - Deprecated

GPT-4o

GPT-4o Search Preview

GPT-4o mini

GPT-4.1

GPT-4.1-mini

GPT-4.1-nano

GPT-4o mini Search Preview

O1

O1 mini

GPT-4 Turbo

GPT-3.5 Turbo

Open-Source Models

GPT-OSS-120B

GPT-OSS-120B (AWS Bedrock)

GPT-OSS-20B

Key Features of Open-Source Models

Video Generation Models

Sora 2

Sora 2 Pro

Image Generation Models

GPT-Image-2

GPT Image 2 Cost Calculation for Image Edits

GPT-Image-1

GPT-Image-1.5

GPT-Image-1-Mini

Embedding Models

text-embedding-3-large

text-embedding-3-small

GPT-4o Transcribe & GPT-4o mini Transcribe

GPT-4o Transcribe Diarize

text-embedding-ada-002

Audio Models

GPT-Audio-1.5

GPT Audio

GPT Audio Mini

Whisper

GPT-4o mini TTS

TTS (Text-to-Speech)

Web Search Capabilities

Types of Web Search

Web Search Tool Configuration

Basic Web Search

Domain Filtering

User Location

Sources Field

Response Format and Citations

Example Response Structure

API Compatibility