Skip to content
VECTOR WIREAI INTELLIGENCE
⌘K
UTC
Models787Deals30d246Policy30d64, 13 pendingSources1,077

GPT-6 Astra

OpenAIReleased Sep 3, 2026Current

Score basis

GPT-6 Astra is at the frontier in reasoning, factuality, and math; strong in long context; capable in agentic tasks, coding, and multimodal tasks; and behind the leaders in instruction following. Too few results yet to rate safety or multilingual tasks.

Capability profile

Bars show the model's median result as a share of the leading model's, per capability. Select a row to see its results.

Frontier3 capabilities
  1. ReasoningLeads6 of 6
  2. FactualityLeads3 of 4
  3. Math−1.4%5 of 5
Strong1 capability
  1. Long Context−10.0%1 of 3
Capable3 capabilities
  1. Agentic−14.8%4 of 7
  2. Coding−20.8%5 of 10
  3. Multimodal−21.9%1 of 6
Limited1 capability
  1. Instruction Following−30.1%1 of 3
Not rated2 capabilities

Too few results yet to rate Safety or Multilingual.

Price

$10.00input$50.00outputper million tokens

From Artificial Analysis · 3 providers tracked · All prices

Costs more than 95% of 332 priced models · 3:1 input-to-output blend, log scale

Evidence

188results on73benchmarks

  • 44 independently verified
  • 94 aggregator
  • 24 vendor-reported
  • 26 cross-referenced

From 24 sources · latest Oct 20, 2026 · How verification works

API features

Tool callingStructured outputsReasoning

As listed by OpenRouter

Research

94 papers reference GPT-6 Astra

GPT-6 Astra benchmark results

188 results on 73 benchmarks, grouped by capability. Each bar is the result as a share of the capability leader's; every result links to its source.

Reasoning

FrontierFull reasoning ranking

Leads the field6 of 6 ranked benchmarks measured

Show 35 more reasoning resultsHide 35 reasoning results

Factuality

FrontierFull factuality ranking

Leads the field3 of 4 ranked benchmarks measured

Show 11 more factuality resultsHide 11 factuality results

Math

FrontierFull math ranking

1.4% behind the leader5 of 5 ranked benchmarks measured

Show 4 more math resultsHide 4 math results

Long Context

StrongFull long context ranking

10.0% behind the leader1 of 3 ranked benchmarks measured

Show 3 more long context resultsHide 3 long context results

Agentic

CapableFull agentic ranking

14.8% behind the leader4 of 7 ranked benchmarks measured

Show 17 more agentic resultsHide 17 agentic results

Coding

CapableFull coding ranking

20.8% behind the leader5 of 10 ranked benchmarks measured

Show 9 more coding resultsHide 9 coding results

Multimodal

CapableFull multimodal ranking

21.9% behind the leader1 of 6 ranked benchmarks measured

Show 6 more multimodal resultsHide 6 multimodal results

Instruction Following

LimitedFull instruction following ranking

30.1% behind the leader1 of 3 ranked benchmarks measured

More results

Benchmarks outside the capability baskets. They are not ranked against a leader.

Show 74 more resultsHide 74 results

GPT-6 Astra: common questions

Who makes GPT-6 Astra?

GPT-6 Astra is made by OpenAI.

When was GPT-6 Astra released?

GPT-6 Astra was released on Sep 3, 2026, according to Artificial Analysis.

What is GPT-6 Astra good at?

GPT-6 Astra is at the frontier in reasoning, factuality, and math; strong in long context; capable in agentic tasks, coding, and multimodal tasks; and behind the leaders in instruction following. Too few results yet to rate safety or multilingual tasks.

How much does GPT-6 Astra cost?

GPT-6 Astra costs $10.00 per million input tokens and $50.00 per million output tokens, according to Artificial Analysis. We track its price at 3 providers. At a mix of three input tokens to one output token, it costs more than 95% of the 332 priced models we track.

How many benchmarks has GPT-6 Astra been tested on?

We track 188 results for GPT-6 Astra on 73 benchmarks from 24 sources, 44 of them independently verified. The latest was recorded on Oct 20, 2026.

Which API features does GPT-6 Astra support?

OpenRouter lists tool calling, structured outputs, and reasoning for GPT-6 Astra.

About this record

Where GPT-6 Astra's numbers come from, and every name it appears under.

Tracked since
Aug 25, 2026
Newest source mention
Oct 13, 2026

Where the results come from

Verification: 188 scores · 44 independently verified · 94 aggregator-attributed · 26 vendor cross-reference · 24 vendor-reported. How these tiers are assigned

From 24 sources on 14 sites. Artificial Analysis supplies 95 of them; the 44 independently verified results come from 9 sites. Bars are coloured by trust tier.

  • artificialanalysis.ai95
  • arcprize.org21
  • deepmind.google20
  • api.llm-stats.com12
  • deploymentsafety.openai.com10
  • livebench.ai7
  • epoch.ai6
  • www-cdn.anthropic.com5
  • raw.githubusercontent.com4
  • anthropic.com3
  • datasets-server.huggingface.co2
  • labs.scale.com1
  • lmarena.ai1
  • simple-bench.com1

Also known as

How our sources name GPT-6 Astra at each reasoning setting.

SettingShort formAPI id
lowgpt-6 astra - provider adapter (low) gpt-6 astra (low)gpt-6-astra-low
mediumgpt-6 astra (medium) gpt-6 astra - provider adapter (medium)gpt-6-astra-medium
highgpt-6 astra (high) gpt-6 astra - provider adapter (high) gpt-6 astra extra highgpt-6-astra-high gpt-6.0-astra-high
xhighgpt-6 astra - provider adapter (xhigh) gpt-6 astra (xhigh)gpt-6-astra-xhigh
maxgpt-6 astra - provider adapter (max) gpt-6 astra (max) gpt 6 astra max gpt6 astra max—
Also listed asgpt-6 astra (non-reasoning)gpt-6 astra - provider adapter (none)openai gpt-6 astragpt-6-astra-non-reasoninggpt-6 "astra"gpt-6 astra (none)gpt-6 astra progpt-6-astra:batchgpt-6-astra-pro:batchgpt6 astrachatgpt-6 astragpt-6astragpt–6 astragpt-6-astra (pro)gpt six astra