Choose Auto, TokenLab Verified, or Official for each request, with prices shown up front. See what's new

OpenAI: GPT-6.2 Astra

GPT-6.2 Astra is OpenAI's most capable GPT-6.2 model for demanding reasoning and coding, with a 1.05M-token context window and configurable reasoning.
Compare models
gpt-6.2-astra
AvailableOpenAIChatCached input 95% less
Input / Output-70%
$8.00 / $40.00$2.40 / $12.00
Context
922K
Released
Nov 17, 2026
Max output
128K
Modalities
VisionChat
Capabilities
Tool usePrompt CacheReasoning

About GPT-6.2 Astra

GPT-6.2 Astra is OpenAI's most capable GPT-6.2 model, built for demanding reasoning and coding projects that run from start to finish. OpenAI says it can build and revise software, use a computer, and work with documents. It takes text and images, replies in text, and lets you set reasoning effort per request. Against GPT-6 Astra, OpenAI reports a higher score on Agents' Last Exam and a later knowledge cutoff.

Where it works well

  • On Agents' Last Exam, OpenAI reports GPT-6.2 Astra at xhigh effort scoring 5.2 points above GPT-6 Astra at max effort.
  • OpenAI says it can build and revise software, use a computer, and work with documents, citing slideshows and websites as examples.
  • Accepts images alongside text, so a screenshot or diagram can travel with the instruction.
  • Reasoning effort is configurable from low to max, and OpenAI suggests medium as the starting point for a new integration.
  • OpenAI's GPT-6.2 guide says its writing is more concise and direct by default than earlier models.

When to choose another model

  • OpenAI recommends Sol for complex work at lower cost, so Astra is the heavier choice for routine coding and everyday requests.
  • The API rejects requests that set reasoning effort to none, so even simple calls involve some thinking.
  • OpenAI launched it with tighter safeguards for higher-risk cybersecurity assistance, so some security requests may be refused.

Getting started

  1. Create API key

    Create a key in Console, then use it with every model on the platform.

  2. Send your first request

    Copy the example for your language and run it against the endpoint.

    Text to textResponses API
    POST/v1/responses
    API format:
    curl https://api.tokenlab.sh/v1/responses \
      -H "Content-Type: application/json" \
      -H "Authorization: Bearer sk-xxx" \
      -d '{
        "model": "gpt-6.2-astra",
        "input": "Hello!"
      }'

Pricing

The Verified price applies to Verified, which costs less on most models. The Official price is the model maker's published price and applies to the more reliable Official route. Auto bills the route that completes the request.

Input tokens <= 272K

per 1M tokens
Official price
Input $8.00 / Output $40.00 / Cache read $0.40 / Cache write $10.00
Verified price
Input $2.40 / Output $12.00 / Cache read $0.12 / Cache write $3.00
Discount
-70%

Input tokens 272K-1.05M

per 1M tokens
Official price
Input $16.00 / Output $60.00 / Cache read $0.80 / Cache write $20.00
Verified price
Input $4.80 / Output $18.00 / Cache read $0.24 / Cache write $6.00
Discount
-70%
Prompt cache pricing

Cache read

Official price
$0.40
Verified price
$0.12
Discount
-70%

Cache write

Official price
$10.00
Verified price
$3.00
Discount
-70%

Usage & activity

Success rate is the share of requests that completed. Latency is how long a full response takes; P95 means 95% of requests finished within that time.

Usage & availability

Last 24 hours
Requests
Success rate
P95 latency
Total tokens
Model performance
Metrics appear once privacy and data volume thresholds are met.

Data is based on aggregate user requests, excluding status checks.

Open in Console

Open GPT-6.2 Astra in Console with a prompt ready to edit or send.

Help me try gpt-6.2-astra with a short message at /v1/responses. Show the reply, latency, and cost.

Use cases

Best for
  • Code
  • Reasoning
  • Vision
  • Trace and fix a bug across services

    Give it the failing workflow, ask it to find the cause, make the fix and run the affected tests, then review the change before merging.

  • Build a deck from a report

    Provide the report and your slide template, and ask for the findings in your team's layout and writing style. Check every figure against the source before sending.

  • Investigate a research question

    Supply the sources and ask it to compare their findings and point out where the evidence disagrees, then verify its conclusions against the data.

Prompt examples

Checkout fails when a coupon expires mid-session. Trace the cause across the services, fix it, run the affected tests, and summarize what changed.

Here is the quarterly report and our slide template. Build a ten-slide deck in the template's layout and list every figure you took from the report.

Compare what these three papers conclude about the same question, say where their evidence disagrees, and state which claims the data supports.

This model has conditional pricing. Monthly token totals alone cannot produce a reliable estimate; use the detailed pricing for the request specification, cache, and applicable time window.

FAQ

What is GPT-6.2 Astra?

GPT-6.2 Astra is OpenAI's most capable GPT-6.2 model, aimed at hard projects that run from start to finish, such as software work, computer use and documents. GPT-6.2 Sol is the everyday model beside it, and GPT-6 Astra is its predecessor.

How does GPT-6.2 Astra differ from GPT-6 Astra?

OpenAI reports that on Agents' Last Exam, GPT-6.2 Astra at xhigh effort scores 5.2 points above GPT-6 Astra at max effort. Its model page lists a May 2026 knowledge cutoff, against April 2026 for GPT-6 Astra. Test both on your own tasks before switching.

When should I choose Sol instead of Astra?

OpenAI recommends Astra for the most demanding tasks and Sol for complex work at lower cost. If Sol passes your own evaluations on everyday coding or agent work, it is the cheaper default, and Astra can take the hardest projects.

What changes when I move a request to GPT-6.2?

OpenAI's guide says to omit temperature and top_p, and to omit logprobs in Chat Completions. It suggests starting at medium reasoning effort, or low if you previously used none, because none is rejected.

Can GPT-6.2 Astra call tools and accept images?

Yes to both. It accepts text and images and returns text. OpenAI's guide says Chat Completions is supported but tool calling requires the Responses API. Audio and video are not supported.

How much does GPT-6.2 Astra cost?

On TokenLab, GPT-6.2 Astra costs Input $2.40 / Output $12.00 per 1M tokens. The pricing table above shows the full breakdown. Rates depend on the billing unit, specification, and usage. Compare matching conditions in the model's detailed pricing; a single rate does not determine the total cost.

What are the context window and output limit of GPT-6.2 Astra?

GPT-6.2 Astra accepts up to 922,000 tokens of context and returns up to 128,000 tokens in one response.

Which endpoint should GPT-6.2 Astra use?

Use https://api.tokenlab.sh/v1/responses for GPT-6.2 Astra. The request example below shows the matching code shape.

Which operations does GPT-6.2 Astra support?

GPT-6.2 Astra supports Text to text. Select an operation above to see its endpoint and request example.

Compare GPT-6.2 Astra

Sources

Reviewed Nov 19, 2026

More from GPT-6

Related models