APIMODELSAPIMODELS
All
Models/GPT-6.2 Astra
OpenAI

GPT-6.2 Astra API

gpt-6.2-astra

GPT-6.2 Astra is OpenAI’s frontier model (Nov 2026): 6.2 points above GPT-6 Astra on GDPval v2 with 22% fewer output tokens and 5.5 above Claude Fable 5.2 on Agents’ Last Exam, per OpenAI. It outputs text only, rejects reasoning_effort none, and bills inputs over 272K tokens at 2×/1.5×. Call gpt-6.2-astra at $4 / $20 per 1M tokens.

50% Below OfficialNewer than GPT-6 Astra1.05M ContextReasoning Effort
Input$4.00
Output$20.00
per 1M tokens

API Documentation

View complete API reference with all parameters and examples.

View Docs

Advanced Features

Streaming

Enable real-time streaming responses with Server-Sent Events.

{
  "model": "gpt-6.2-astra",
  "stream": true,
  "messages": [...]
}

Function Calling (Tools)

Enable the model to use tools and call functions.

{
  "model": "gpt-6.2-astra",
  "tools": [{
    "type": "function",
    "function": {
      "name": "get_weather",
      "description": "Get current weather for a location",
      "parameters": {
        "type": "object",
        "properties": {
          "location": {"type": "string", "description": "City name"}
        },
        "required": ["location"]
      }
    }
  }],
  "messages": [{"role": "user", "content": "What's the weather in Tokyo?"}]
}

JSON Mode

Get structured JSON responses from the model.

{
  "model": "gpt-6.2-astra",
  "response_format": {"type": "json_object"},
  "messages": [{"role": "user", "content": "Extract info as JSON: John is 30 years old"}]
}

API Parameters Reference

ParameterTypeRequiredDescription
modelstringYesModel identifier (e.g., gpt-6.2-astra)
messagesarrayYesArray of message objects with role and content
max_tokensintegerNoMaximum tokens in the response
streambooleanNoEnable streaming responses (SSE)
temperaturenumberNoSampling temperature (0.0 - 2.0)
top_pnumberNoNucleus sampling threshold (0.0 - 1.0)
toolsarrayNoFunction calling tools definition
response_formatobjectNoOutput format (e.g., json_object)

Full API Documentation

View complete API reference with streaming, thinking, and more.

View Documentation

Pricing

Input
$4.00
per 1M tokens
Output
$20.00
per 1M tokens

Billing: Cost = (input_tokens * input_price + output_tokens * output_price) / 1,000,000

50% Below OpenAI List

$4.00 / $20.00 per 1M tokens against the $8 / $40 list; cached input $0.20

Ahead of GPT-6 Astra on Long Work

OpenAI: GDPval v2 6.2 points higher with 22% fewer output tokens; Terminal-Bench-Science 0.1 9 points higher

Ahead of Fable 5.2 on OpenAI's Tests

OpenAI: 5.5 points above Claude Fable 5.2 on Agents' Last Exam at 5% of its cost; first on AA-3D at 1,318 Elo

1.05M Context, 128K Output

Text and image input, text output only; knowledge cutoff 18 May 2026; >272K bills input 2× and output 1.5×

Always Reasons

reasoning_effort none is rejected; reasoning tokens bill as output and count against max_tokens

TL;DR GPT-6.2 Astra is an OpenAI large language model, callable via APIMODELS' unified API (model name `gpt-6.2-astra`). Pricing: Input: $4.00, Output: $20.00 per 1M tokens. One API key for all image / video / LLM / audio models.

About GPT-6.2 Astra

GPT-6.2 Astra is a Large Language Model API provided by OpenAI. GPT-6.2 Astra is the frontier tier of OpenAI's GPT-6.2 release, launched on 17 November 2026 together with GPT-6.2 Sol; OpenAI calls it its most capable model for your hardest projects, from start to finish, able to build and revise software, use a computer and work with documents. On APIMODELS it costs $4.00 per 1M input tokens and $20.00 per 1M output — half OpenAI's $8 / $40 list — with cached input at $0.20. OpenAI cut Astra's list price by 20% from GPT-6 Astra's $10 / $50 and its cached rate by 60%; on APIMODELS, though, GPT-6 Astra sells at $3.40 / $17.00, so per token 6.2 Astra costs more here and any saving has to come from the fewer tokens OpenAI says it uses. OpenAI's launch figures: on GDPval v2, its benchmark of long professional workflows, GPT-6.2 Astra leads the models OpenAI tested, 6.2 points above GPT-6 Astra with 22% fewer output tokens; on Agents' Last Exam it is 5.2 points above GPT-6 Astra at under a third of the cost and 5.5 points above Claude Fable 5.2 at 5% of its cost; on Terminal-Bench-Science 0.1 it is 9 points above GPT-6 Astra at under a third of the cost; on Harvey's LAB-AA v1.1 legal benchmark its pass rate is more than double Fable 5.2's at about one sixth of the cost; and in Artificial Analysis's blind AA-3D vote it ranks first at 1,318 Elo, ahead of Fable 5.2 (1,271) and GPT-6.2 Sol (1,259). OpenAI also says its slides, spreadsheets and documents follow your templates more closely and read less like AI writing, and that its code is cleaner. These are OpenAI's own numbers, several measured on OpenAI's internal configurations. The specifications: a 1,050,000-token context window with up to 922,000 input tokens, 128,000 max output, text and image input with text output only, and a knowledge cutoff of 18 May 2026. Limits: the API rejects reasoning_effort none; when a request's input exceeds 272K tokens, input and cached input bill at 2× and output at 1.5× for the whole request; OpenAI ships it with stricter safeguards on higher-risk cybersecurity requests; and the computer use OpenAI demonstrates relies on its hosted tools — OpenAI's hosted Responses tools have not been verified on this endpoint (GPT-6 Astra's computer_use_preview and file_search are not available here), so build agent loops on function calling. Call it through the OpenAI-compatible /v1/chat/completions or /v1/responses with model gpt-6.2-astra; the OpenAI SDK only needs base_url and key changed. Streaming and function calling work, including parallel tool calls. OpenAI's Batch API and its Fast and Ultrafast modes are not offered here; every request bills at the standard rates above, and failed requests are not charged. Through APIMODELS platform, you can access this model via a unified API with transparent pay-as-you-go pricing. Current pricing: Input: $4.00, Output: $20.00 per 1M tokens.

Key Features

  • 50% Below OpenAI List -- $4.00 / $20.00 per 1M tokens against the $8 / $40 list; cached input $0.20
  • Ahead of GPT-6 Astra on Long Work -- OpenAI: GDPval v2 6.2 points higher with 22% fewer output tokens; Terminal-Bench-Science 0.1 9 points higher
  • Ahead of Fable 5.2 on OpenAI's Tests -- OpenAI: 5.5 points above Claude Fable 5.2 on Agents' Last Exam at 5% of its cost; first on AA-3D at 1,318 Elo
  • 1.05M Context, 128K Output -- Text and image input, text output only; knowledge cutoff 18 May 2026; >272K bills input 2× and output 1.5×
  • Always Reasons -- reasoning_effort none is rejected; reasoning tokens bill as output and count against max_tokens

Use Cases

Chatbot & Customer Support

Build intelligent conversational systems to automatically answer user queries and improve service efficiency.

Content Generation

Automatically write articles, emails, ad copy, and other text content to boost productivity.

Code Assistant

Assist with code writing, debugging, and code review to accelerate software development.

Data Analysis

Understand and analyze unstructured data, extract key insights, and generate summary reports.

Why APIMODELS

  • Unified API -- One API key to access all models, no need to register on multiple platforms
  • Transparent Pricing -- Pay-as-you-go with per-call pricing published on every model page, ideal for indie developers and startups
  • Instant Access -- Start right after signup; pay by card, PayPal, Alipay or USDT
  • Full Documentation -- Detailed API docs with code examples in cURL, Python, and Node.js

Frequently Asked Questions

How to use GPT-6.2 Astra API?

Sign up at APIMODELS, get your API key, and call our unified API endpoint. We provide detailed API documentation with code examples in cURL, Python, and Node.js.

What is the difference between APIMODELS and the official OpenAI API?

APIMODELS offers the same GPT-6.2 Astra model through our aggregation platform. We provide a unified API interface so you do not need separate accounts for each provider - one API key to access all models.

What is GPT-6.2 Astra API pricing?

$4.00 per 1M input tokens, $20.00 per 1M output tokens and $0.20 per 1M cached input on APIMODELS — half OpenAI's $8 / $40 / $0.40 list. OpenAI set that list 20% below GPT-6 Astra's $10 / $50 on input and output and cut the cached rate by 60%. One thing to know when comparing on this site: GPT-6 Astra is $3.40 / $17.00 here, so per token 6.2 Astra is the pricier of the two; OpenAI's case is that it finishes tasks with fewer tokens. Past 272K input tokens a request bills at 2× input and 1.5× output, reasoning tokens bill as output, and failed requests are not charged.

OpenAI: GPT-6.2 Astra model page ↗

How much does a GPT-6.2 Astra task cost?

It depends on how many tokens the task uses, and Astra reasons on every request, so output usually dominates: $20.00 per 1M output tokens here, $4.00 per 1M input and $0.20 per 1M cached input. OpenAI's argument is cost per task rather than per token: it reports 22% fewer output tokens than GPT-6 Astra on GDPval v2, and under a third of GPT-6 Astra's cost on Agents' Last Exam and Terminal-Bench-Science. To keep bills down, stay under 272K input tokens per request (beyond that the whole request bills at 2× input and 1.5× output), keep a stable prompt prefix so repeated context bills at the cached rate, and send routine steps to GPT-6.2 Sol at $1.00 / $5.00.

When was GPT-6.2 Astra released?

OpenAI released GPT-6.2 Astra on 17 November 2026, together with GPT-6.2 Sol, in ChatGPT (Plus, Pro, Enterprise and Edu), Codex and the OpenAI API under the id gpt-6.2-astra. It has been available on APIMODELS since 18 November 2026. It succeeds GPT-6 Astra from September 2026 — there was no 6.1 Astra — and GPT-6 Astra stays available here.

What are the GPT-6.2 Astra benchmarks?

From OpenAI's launch post and system card of 17 November 2026: GDPval v2 — first among the models OpenAI tested, 6.2 points above GPT-6 Astra with 22% fewer output tokens; Agents' Last Exam — 5.2 points above GPT-6 Astra at under a third of the cost, and 5.5 points above Claude Fable 5.2 at 5% of its cost; Terminal-Bench-Science 0.1 — 9 points above GPT-6 Astra at under a third of the cost; Harvey LAB-AA v1.1 — a pass rate more than double Fable 5.2's at about one sixth of the cost; AA-AnalystAgent — 62.5% of questions answered correctly on all five attempts; AA-3D — first at 1,318 Elo; GPQA Diamond without tools — 97.0%. All are OpenAI's numbers, and OpenAI notes that Fable 5.2's Agents' Last Exam result excludes nine tasks that used a fallback model.

OpenAI: Introducing GPT-6.2 Astra and Sol ↗

How much better is GPT-6.2 Astra than GPT-6 Astra?

On OpenAI's numbers, a few points better and clearly more efficient: 6.2 points higher on GDPval v2 with 22% fewer output tokens, 5.2 points higher on Agents' Last Exam and 9 points higher on Terminal-Bench-Science 0.1, each at under a third of GPT-6 Astra's cost, and 97.0% against 96.0% on GPQA Diamond. Partners quoted by OpenAI report the same pattern — Harvey saw more than three times as many Legal Agent Benchmark tasks completed with 34.5% fewer output tokens. Two caveats: these are vendor results, not independent tests, and on this site GPT-6 Astra is $3.40 / $17.00 per 1M tokens against 6.2 Astra's $4.00 / $20.00, so the per-task saving depends on your workload really using fewer tokens.

Is GPT-6.2 Astra better than Claude Fable 5.2?

On OpenAI's published tests, yes: 5.5 points higher on Agents' Last Exam at 5% of Fable 5.2's cost, more than double its pass rate on Harvey LAB-AA v1.1 at about a sixth of the cost, and 1,318 against 1,271 Elo in the AA-3D blind vote. Those results come from OpenAI, and on Agents' Last Exam Fable 5.2's score excludes nine tasks that used a fallback model; Anthropic's own table puts Fable 5.2 at 74.6% on Terminal-Bench-Science 0.1 in Claude Code, a different setup from the one OpenAI charts. Per-token prices are close here — $4.00 / $20.00 for Astra, $4.80 / $24.00 for Fable 5.2 — so run both on your own tasks; the comparison page sets the numbers side by side.

GPT-6.2 Astra vs Claude Fable 5.2 ↗

GPT-6.2 Astra vs Claude Opus 5.5: which should I use?

They sit at different price points: Opus 5.5 is $2.40 / $12.00 per 1M tokens here and GPT-6.2 Astra is $4.00 / $20.00, so Astra has to finish tasks in fewer tokens or with clearly better results to pay off. OpenAI's launch post does not compare Astra with Opus 5.5, and Anthropic positions Opus 5.5 as reaching Fable 5.1-level results on most work at a lower running cost. Choose by integration and task: Opus 5.5 runs natively on /v1/messages for Claude Code; Astra uses OpenAI's chat/completions and responses formats for Codex-style tools, and OpenAI's strongest claims for it are in long professional work — documents, spreadsheets, science and legal. One APIMODELS key calls both, so test each on a sample of your workload.

GPT-6.2 Astra vs GPT-6.2 Sol: when is Astra worth four times the price?

When the task is long, professional and judged on the finished deliverable. Per token Astra costs four times Sol: $4.00 / $20.00 against $1.00 / $5.00 here. OpenAI's numbers put Sol within 0.5 points of Astra on FrontierCode 2.0 at about a fifth of the cost, so for most coding Sol is the better buy; Astra pulls ahead on GDPval v2, on AA-3D (1,318 against 1,259 Elo) and on the science, legal and spreadsheet work OpenAI trained it for. A common setup is Sol for routine steps and Astra for planning or final drafts — both ids work on the same key.

What's new in GPT-6.2 Astra?

OpenAI's list: slides, spreadsheets and documents that read less like AI writing and follow your templates down to font sizes; cleaner code with more targeted tests and fewer review rounds; front ends that look and work better, especially websites, games and 3D; more precise computer use in CAD and simulation tools; better judgment about which tool to use; fewer output tokens for the same work; and a list price 20% below GPT-6 Astra's, with cached input 60% lower. It is also the first OpenAI model released externally to reach the High threshold for AI self-improvement in its Preparedness Framework, and like Sol it ships with stricter safeguards on high-risk cybersecurity requests.

Which AI models are better than Claude Fable 5?

Anthropic's own successors, to start with: Fable 5.1 scores 52.6% on Terminal-Bench-Science against Fable 5's 24.7%, and Fable 5.2 scores 74.6%. On OpenAI's published tests, GPT-6.2 Astra goes further than Fable 5.2 on Agents' Last Exam (5.5 points higher) and Harvey LAB-AA (more than double the pass rate). Prices here: Fable 5 is $5 / $25 per 1M tokens, Fable 5.2 $4.80 / $24.00 and GPT-6.2 Astra $4.00 / $20.00, so moving a Fable 5 project to a newer model does not cost more per token.

What can GPT-6.2 Astra do?

GPT-6.2 Astra supports: 50% Below Official, Newer than GPT-6 Astra, 1.05M Context, Reasoning Effort. See the APIMODELS docs for full parameters and call examples.

Can I use the GPT-6.2 Astra API from any country, including mainland China?

Yes. APIMODELS does not restrict access by country: the site and the API answer requests from anywhere, and apimodels.app and api.apimodels.app are reachable directly from mainland China without a proxy. You call GPT-6.2 Astra with one APIMODELS key, so you do not need a separate account with the model's developer.

What payment methods are supported? Can I pay with PayPal or crypto?

Yes to both. You can top up with a card (Visa, Mastercard and others, through Stripe), PayPal, Alipay, or USDT on TRON (TRC20) or BNB Smart Chain (BEP20). Card, PayPal and Alipay top-ups are credited immediately; USDT is credited once the transfer confirms on-chain. Auto top-up is optional, and a $50 top-up adds a $1 bonus ($3 on a $100 top-up). The balance is in USD and billed per use, with no subscription.

Related models

Related reading

How to get access, regional availability, and how this model compares with its alternatives.