New Flagship Model Added: GPT-6 Astra

Date: 2026-09-05 / (1405-06-14)

Summary

OpenAI's new flagship model, GPT-6 Astra, is now available on AvalAI as gpt-6-astra. It is designed for demanding computer-use, software-engineering, professional, scientific, mathematical, cybersecurity, and long-context workloads. AvalAI provides full support through v1/chat/completions, v1/messages, and v1/responses.

---

Details

OpenAI

GPT-6 Astra is OpenAI's new-generation flagship model. OpenAI reports improvements across browser and computer use, coding, professional knowledge work, scientific analysis, cybersecurity, long-context retrieval, and adherence to task boundaries.

Key features:

• Computer and browser use: Designed to operate interfaces, gather information, and complete multi-step tasks
• Software engineering: Suited to repository-scale analysis, implementation, debugging, and verification
• Professional work: Handles research, document analysis, quantitative work, and structured deliverables
• Science and mathematics: OpenAI reports stronger performance on advanced scientific and abstract-reasoning evaluations
• Cybersecurity: Supports defensive analysis and authorized security workflows with improved safeguards
• Long-context retrieval: Designed to locate and use relevant evidence across large context windows
• Task-boundary alignment: Improved adherence to user intent, scope, and operational constraints
• Reasoning effort: Supports reasoning controls for balancing response quality, latency, and cost
• Endpoint support: Full support on v1/chat/completions, v1/messages, and v1/responses

Endpoint Availability

Endpoint	Support	Notes
v1/chat/completions	Full	OpenAI-compatible chat, reasoning, structured output, and tool use
v1/messages	Full	Anthropic-compatible messages and tool workflows
v1/responses	Full	Responses-first reasoning, stateful workflows, and supported tools

---

Pricing

Prices are in USD per 1 million tokens. Requests with more than 272,000 input tokens use the long-context rates shown below.

Token Category	Up to 272K Input	Above 272K Input
Input	$10.00	$20.00
Cached input	$1.00	$2.00
Cache creation input	$12.50	$25.00
Output	$50.00	$75.00

Only requests above the 272K input threshold use the higher rates. Actual cost depends on uncached input, eligible cache reads and writes, generated output, and the request's total input length.

---

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": "gpt-6-astra",
    "messages": [
      {
        "role": "user",
        "content": "Review this incident timeline, identify the most likely root cause, and propose a verification plan."
      }
    ]
  }'

Response

The shortened response below illustrates the standard Chat Completions structure. Token counts and cost are illustrative and vary with the request.

{
  "id": "chatcmpl-gpt-6-astra-example",
  "created": 1788624000,
  "model": "gpt-6-astra",
  "object": "chat.completion",
  "system_fingerprint": null,
  "choices": [
    {
      "finish_reason": "stop",
      "index": 0,
      "message": {
        "content": "The strongest hypothesis is a race between lease expiry and retry processing. Verify it by correlating duplicate job IDs with lease timestamps, reproducing the timing under load, and confirming that an idempotency guard prevents duplicate writes.",
        "role": "assistant",
        "annotations": []
      }
    }
  ],
  "usage": {
    "completion_tokens": 70,
    "prompt_tokens": 25,
    "total_tokens": 95,
    "completion_tokens_details": null,
    "prompt_tokens_details": {
      "audio_tokens": null,
      "cached_tokens": 0,
      "text_tokens": 25,
      "image_tokens": null
    }
  },
  "estimated_cost": {
    "unit": "0.0037500000",
    "irt": 574.5,
    "exchange_rate": 153200
  }
}

---

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": "gpt-6-astra",
    "messages": [
      {
        "role": "user",
        "content": "Analyze this repository architecture and propose a verified modernization plan."
      }
    ]
  }'

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="gpt-6-astra",
    messages=[
        {
            "role": "user",
            "content": "Analyze this repository architecture and propose a verified modernization plan.",
        }
    ],
)

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: "gpt-6-astra",
  messages: [
    {
      role: "user",
      content: "Analyze this repository architecture and propose a verified modernization plan.",
    },
  ],
});

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

Responses API

Use the Responses API for Responses-first reasoning and stateful workflows:

Bash

curl https://api.avalai.ir/v1/responses \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $AVALAI_API_KEY" \
  -d '{
    "model": "gpt-6-astra",
    "reasoning": {"effort": "high"},
    "input": "Design a staged migration and verify its rollback strategy."
  }'

Python

from openai import OpenAI

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

response = client.responses.create(
    model="gpt-6-astra",
    reasoning={"effort": "high"},
    input="Design a staged migration and verify its rollback strategy.",
)

print(response.output_text)

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.responses.create({
  model: "gpt-6-astra",
  reasoning: { effort: "high" },
  input: "Design a staged migration and verify its rollback strategy.",
});

console.log(response.output_text);

---

Reasoning Effort

Start with medium reasoning effort for general work and use high when evaluations show that additional reasoning improves task success. OpenAI's published evaluations include low-, medium-, and high-effort configurations. Do not ask the model to expose hidden chain-of-thought; request a concise rationale, supporting evidence, or a verification checklist instead.

---

Recommended Workloads

Workload	Why GPT-6 Astra Fits
Computer and browser use	Designed for multi-step interaction with software interfaces and web environments
Complex software engineering	Supports repository analysis, implementation, debugging, and verification
Professional knowledge work	Produces structured research, analysis, and quantitative deliverables
Science and mathematics	OpenAI reports strong results on advanced scientific and reasoning evaluations
Authorized security analysis	Supports defensive cybersecurity investigation with improved safeguards
Long-context tasks	Retrieves and applies evidence from extensive documents and codebases

---

Related Links

• OpenAI Models Overview
• Reasoning Models Guide
• Chat Completions API Reference
• Messages API Reference
• Responses API Reference
• Pricing