New Models Added: Gemini 3.6 Flash and Gemini 3.5 Flash-Lite

Date: 2026-07-21 / (1405-04-30)

Summary

We announce the addition of Google's gemini-3.6-flash and gemini-3.5-flash-lite. Both are natively multimodal reasoning models with 1,048,576-token input windows and are available through the native Gemini v1beta/, v1/chat/completions, v1/messages, and partially through v1/responses.

---

Details

Google Gemini

Gemini 3.6 Flash

gemini-3.6-flash is Google's workhorse model for sustained, high-quality intelligence at Flash speed and cost. It improves coding, knowledge work, multimodal analysis, agentic execution, and spatial reasoning while using fewer output tokens and fewer tool calls than Gemini 3.5 Flash on evaluated workflows.

Key Features:

• Long Context: 1,048,576 input tokens and up to 65,536 output tokens
• Multimodal Input: Text, image, video, audio, and PDF input with text output
• Agentic Execution: Strong performance in coding loops, tool use, multi-step workflows, and computer use
• Token Efficiency: Uses 17% fewer output tokens than Gemini 3.5 Flash on the Artificial Analysis Index
• Developer Features: Thinking, function calling, structured outputs, code execution, prompt caching, file search, Google Search grounding, Google Maps grounding, and URL context
• Computer Use: Supported in preview through the native Gemini tool interface
• Best For: Agentic coding, code migrations, document and chart analysis, knowledge work, and complex multimodal workflows

Gemini 3.5 Flash-Lite

gemini-3.5-flash-lite is Google's fastest and most cost-effective Gemini 3.5-class model. It is optimized for high-throughput agentic workflows, low-latency subagent tasks, document parsing, simple data extraction, classification, translation, and other production workloads where speed and cost are primary constraints.

Key Features:

• Long Context: 1,048,576 input tokens and up to 65,536 output tokens
• Multimodal Input: Text, image, video, audio, and PDF input with text output
• High Throughput: Measured at 350 output tokens per second by Artificial Analysis
• Agentic Efficiency: Designed for high-volume subagent workloads and multi-step execution
• Developer Features: Thinking, function calling, structured outputs, code execution, prompt caching, file search, Google Search grounding, Google Maps grounding, and URL context
• Best For: Document processing, data extraction, routing, classification, search subagents, translation, and cost-sensitive production traffic

Endpoint Availability

Endpoint	Gemini 3.6 Flash	Gemini 3.5 Flash-Lite	Notes
v1beta/	Supported	Supported	Native Gemini request and response schema
v1/chat/completions	Supported	Supported	OpenAI-compatible Chat Completions
v1/messages	Supported	Supported	Anthropic-compatible Messages API
v1/responses	Partial support	Partial support	Verify required parameters and tools before production use

---

Pricing

Prices are in USD per 1 million tokens.

Model	Input	Cached Input	Output
gemini-3.6-flash	$1.50	$0.15	$7.50
gemini-3.5-flash-lite	$0.30	$0.03	$2.50

Gemini 3.6 Flash is suited to workloads that prioritize stronger coding, agentic execution, and knowledge-work quality. Gemini 3.5 Flash-Lite provides the lower-cost option for latency-sensitive and high-volume processing.

---

API Request and Response Example

Request

curl https://api.avalai.ir/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $AVALAI_API_KEY" \
  -d '{
    "model": "gemini-3.6-flash",
    "messages": [
      {
        "role": "user",
        "content": "Review this service architecture and propose a reliable migration plan to event-driven microservices."
      }
    ]
  }'

Response

The following shortened response shows the standard Chat Completions structure:

{
  "id": "chatcmpl-gemini36-example",
  "created": 1784650800,
  "model": "gemini-3.6-flash",
  "object": "chat.completion",
  "system_fingerprint": null,
  "choices": [
    {
      "finish_reason": "stop",
      "index": 0,
      "message": {
        "content": "Start with bounded-context discovery, introduce a transactional outbox, and migrate one low-risk domain at a time behind stable contracts. Add schema versioning, idempotent consumers, distributed tracing, and rollback criteria before shifting production traffic.",
        "role": "assistant",
        "thinking_blocks": [],
        "annotations": []
      }
    }
  ],
  "usage": {
    "completion_tokens": 147,
    "prompt_tokens": 22,
    "total_tokens": 169,
    "completion_tokens_details": null,
    "prompt_tokens_details": {
      "audio_tokens": null,
      "cached_tokens": null,
      "text_tokens": 22,
      "image_tokens": null
    }
  },
  "estimated_cost": {
    "unit": "0.0011355000",
    "irt": 130.13,
    "exchange_rate": 114600
  }
}

---

SDK Usage Examples

Bash

curl https://api.avalai.ir/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $AVALAI_API_KEY" \
  -d '{
    "model": "gemini-3.6-flash",
    "messages": [
      {
        "role": "user",
        "content": "Extract the risks, dependencies, and next actions from this project report."
      }
    ]
  }'

Python

from openai import OpenAI

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

response = client.chat.completions.create(
    model="gemini-3.6-flash",
    messages=[
        {
            "role": "user",
            "content": "Extract the risks, dependencies, and next actions from this project report.",
        }
    ],
)

print(response.choices[0].message.content)

JavaScript

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.AVALAI_API_KEY,
  baseURL: "https://api.avalai.ir/v1",
});

const response = await client.chat.completions.create({
  model: "gemini-3.6-flash",
  messages: [
    {
      role: "user",
      content: "Extract the risks, dependencies, and next actions from this project report.",
    },
  ],
});

console.log(response.choices[0].message.content);

Native Gemini API Example

Use the native Gemini endpoint when your application needs Gemini-specific request fields or tools:

curl https://api.avalai.ir/v1beta/models/gemini-3.5-flash-lite:generateContent \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $AVALAI_API_KEY" \
  -d '{
    "contents": [
      {
        "role": "user",
        "parts": [
          {
            "text": "Classify this support request as billing, account, technical, or other: My password reset link expired."
          }
        ]
      }
    ]
  }'

---

Choosing Between the Models

Workload	Recommended Model	Reason
Agentic coding and code migration	gemini-3.6-flash	Stronger coding quality, fewer execution loops, and efficient tool use
Knowledge work and report drafting	gemini-3.6-flash	Improved document, chart, and data analysis
Complex multimodal workflows	gemini-3.6-flash	Stronger multimodal and spatial reasoning
High-volume extraction and classification	gemini-3.5-flash-lite	Lower token cost and high throughput
Search and routing subagents	gemini-3.5-flash-lite	Low latency with configurable reasoning capability
Large-scale document parsing	gemini-3.5-flash-lite	1M-token context at lower input and output prices

---

Related Links

• Google Models Documentation
• Reasoning Models Guide
• Chat Completions API Reference
• Native Gemini API Reference
• Pricing