GPT-6 Astra

Sep 2026
GPT-6 Astra
Compare in
on
$0.82per task?
$10 / $50per 1M in / outcache $1
56 tok/sfirst token in 2.5 s?
1.05Mcontext

Rankingat low effort?

Full ranking →
1Gemini 3.8 Flash95.32Grok 4.694.93Gemini 3.1 Pro Preview94.13Muse Spark 1.394.13GPT-5.6 Sol94.1
···
8Qwen 2.4T A95B93.511GPT-6 Astra93.111Grok 4.593.1

Strong at

S
Coding#11 of 132Reasoning#12 of 334Science#11 of 312Knowledge#6 of 318Vision#5 of 124
A
Agents & work tasks#39 of 175Long context#41 of 273

Alternatives?

ModelScoreCostPer taskWhat changes
CheaperMiniMax M3thinkingOpensame-38%$0.51Context: 1.05M → 524K+2BetterGemini 3.8 Flashhigh+2%+52%*$1.24Context: 1.05M → 1M+2OpenKimi K3maxsame+145%$2.00Context: 1.05M → 1M+4CheaperDeepSeek V4 Pro 0813maxOpensame-18%$0.67Context: 1.05M → 1M+6CheaperGPT-5.6 Solmediumsame-38%$0.50

Benchmarks

in Chat & Knowledge
AA Intelligence Index 4.3.245.8#15
AA Omniscience Index40.6#6
AA Omniscience · Accuracy59.5#6
AA Omniscience · Hallucination Rate46.9#77
Humanity's Last Exam49.2#12
GPQA Diamond93.1#11
AA-LCR (long context) 1.180#41
CritPt (physics)26.3#17
GDP.pdf · AA all-pass30.4#4
ARC-AGI-1 · public evaluation98.3#7
ARC-AGI-1 · semi-private96.5#10
ARC-AGI-2 · public evaluation94.2#7
ARC-AGI-2 · semi-private85.4#10
ARC-AGI-3 · semi-private17.5#4
FrontierMath Tier 4 (v2)87.8#5
MathArena · ArXivMath · August 202685.1#5
MathArena · ArXivMath · June 202684.4#6
MathArena · ArXivMath · May 202688.8#3
MathArena · BrokenArXiv · August 202674.7#5
MathArena · BrokenArXiv · June 202697.7#3
MathArena · BrokenArXiv · May 202688.5#3
AA Cyber Index successeslab33
HealthBench Professional length-adjustedlab70.3
HealthBench length-adjustedlab58.3
HealthBench Hard length-adjustedlab36.6
HealthBench Consensus length-adjustedlab95.5

Watch out 3

Monitoring can return HTTP 403›

Misalignment monitoring can return HTTP 403 before streaming or errors after output; match misalignment_policy_violation and handle stream errors. Source ↗

No none reasoning effort›

The none reasoning effort is unsupported on GPT-6 Astra and GPT-6.1 Sol; use low instead. Source ↗

Remove unsupported sampling parameters›

When reasoning effort is not none, remove temperature, top_p and top_logprobs, plus Chat Completions logprobs or Responses output logprobs inclusion. Source ↗

FAQ

How much does GPT-6 Astra cost?›

$10 per 1M input tokens and $50 per 1M output tokens; cached input $1 per 1M. A typical benchmark task costs about $0.82 at low effort.

Is GPT-6 Astra free to use?›

There is no free API; the cheapest paid price is $10 per 1M input tokens.

How good is GPT-6 Astra, and what is it best at?›

#12 of 334 in Chat & Knowledge, #11 of 132 in Coding, #21 of 133 in Agents & Tool Use, #26 of 195 in Writing & Language, #5 of 124 in Vision & Documents. Strongest at: Coding, Agents & work tasks, Reasoning, Science, Long context, Knowledge, Vision.

Is GPT-6 Astra better than Gemini 3.8 Flash?›

On GPQA Diamond, GPT-6 Astra scores 96.3 vs 95.3 — higher. It is 34% cheaper per task.

How can I use GPT-6 Astra?›

Through the API of Amazon Bedrock, Azure, OpenAI.

What is the context window of GPT-6 Astra?›

1.05M tokens, with up to 128K output tokens.

When was GPT-6 Astra released? Is there a newer version?›

Released 2026-09-04. It is the current version in its line.

Is GPT-6 Astra open source?›

No, GPT-6 Astra is a closed model available through an API. The closest open-weights option here is Kimi K3.

Can GPT-6 Astra read images?›

It accepts text and images, PDF files as input.

What is GPT-6 Astra?›

GPT-6 Astra is a closed language model from OpenAI, released Sep 2026.

What are the best alternatives to GPT-6 Astra?›

MiniMax M3 (cheaper), Gemini 3.8 Flash (better), Kimi K3 (open weights).