GPT-6 Astra
ReleasedModel GPT-6 Astra is a model by OpenAI, released on 3 Sept 2026 (42 days ago).
In the Frontier SetOpenAI positions this model in its top tier — role: Highest capability. “our flagship model for complex reasoning and coding” Source: developers.openai.com · checked on 4 Oct 2026
- Where can it be used?
- Evidenced as usable in: ChatGPT, ChatGPT Work, Codex, GitHub Copilot, Microsoft 365 Copilot (verified for Cowork only)For 9 more recorded products there is no statement; that does not mean “not available”.Available in the provider’s API
- Did we test it?
- No lab run published yet
- Last change
- 9 days ago
- Evidence
- 6 sources from 5 sitesRegistry entry:
What backs this up?
Three kinds of evidence, kept apart. None of them proves AGI on its own; where one is missing, the page says so.
- In real productsEvidenced as usable in: ChatGPT, ChatGPT Work, Codex, GitHub Copilot, Microsoft 365 Copilot (verified for Cowork only)
- agi.jetzt PrüfstandNo published run yet.
- Public benchmarks21 reported results on 17 benchmarks — each with its source
A model and the product around it are different things: a result for one does not carry over to the other.
Public benchmark results
Results reported by benchmark maintainers or providers, each with its source. They are not averaged and not compared across benchmarks; a provider result is self-reported. Some sources are data files of the benchmark’s own site rather than pages; they are marked as machine-readable.
- GPT-6 Astra41.4 %AutomationBench · 1.0.6 · task success ratesource as of 03/10/2026 · benchmark maintainer · model + harness · AutomationBench API mode (47 simulated tools, Pydantic-backed apps) · tools: yes · budget: up to 50 steps per task · reasoning: Max · reported as „GPT 6 Astra (Max)“Source: zapier.com — AutomationBench, GPT-6 Astra
- GPT-6 Astra71 %Chartography · v1 · accuracy (expert-range graded), no toolssource as of 03/10/2026 · benchmark maintainer · mode not stated · judge: google/gemini-3.5-flash · tools: no · setting: Inspect AI (surge-ai/chartography) · attempts: 10 · web: no · reasoning: Max · computer use: no · reported as „GPT 6 Astra (Max reasoning)“Source: surgehq.ai — Chartography, GPT-6 Astra
- GPT-6 Astra34.2 %GDP.pdf · v1 · pass rate, all criteria (strict)source as of 03/10/2026 · benchmark maintainer · mode not stated · judge: google/gemini-3.5-flash · tools: no · setting: Inspect AI (surge-ai/gdp-pdf) · attempts: 5 · web: no · reasoning: Max · computer use: no · reported as „GPT 6 Astra (Max reasoning)“Source: surgehq.ai — GDP.pdf, GPT-6 Astra
- GPT-6 Astra58.18 %Terminal-Bench · 4.0 · resolution rate (mean of 5 runs)source as of 03/10/2026 · benchmark maintainer · model + harness · Codex · tools: yes · attempts: 5 · pass@k: 1 · reasoning: maxSource: www.tbench.ai — Terminal-Bench, GPT-6 Astra
- GPT-6 Astra$15,514.7Vending-Bench 2 · 2 · final bank balancesource as of 03/10/2026 · benchmark maintainer · model + harness · Andon Labs Vending-Bench 2 agent environment (email, web search, inventory tools, notes/reminders) · tools: yes · budget: ~1 simulated year; 3000-6000 messages, 60-100M output tokens per run · attempts: 6 · web: yesMachine-readable source: andonlabs.com · data bundle of the website, not linked
- GPT-6 Astra95 %ARC-AGI-2 · semi-private · accuracy (task pass rate)source as of 02/10/2026 · benchmark maintainer · mode not stated · reasoning: max · reported as „GPT-6 Astra (Max)“Machine-readable source: arcprize.org · JSON — ARC-AGI-2, GPT-6 Astra
- GPT-6 Astra62.71 %ARC-AGI-3 · semi-private · scoresource as of 02/10/2026 · benchmark maintainer · mode not stated · reasoning: max · reported as „GPT-6 Astra (Max)“Machine-readable source: arcprize.org · JSON — ARC-AGI-3, GPT-6 Astra
- GPT-6 Astra98.55 %ARC-AGI-3 · semi-private · score · provider adaptersource as of 02/10/2026 · benchmark maintainer · model + harness · ARC Prize provider adapter · reasoning: max · reported as „GPT-6 Astra - Provider Adapter (Max)“Machine-readable source: arcprize.org · JSON — ARC-AGI-3, GPT-6 Astra
- GPT-6 Astra53.3 %FrontierCode · 1.1-main · overall score (mergeability)source as of 29/09/2026 · benchmark maintainer · mode not stated · judge: model-graded rubric checks (grader not named on leaderboard) · tools: yes · web: yesSource: cognition.com — FrontierCode, GPT-6 Astra
- GPT-6 Astra74.12 %DeepSWE · 1.1 · pass@1 (mean of 4 runs)source as of 22/09/2026 · benchmark maintainer · model + harness · mini-swe-agent · tools: yes · attempts: 4 · pass@k: 1 · reasoning: xhigh · reported as „gpt-6-astra“Source: deepswe.datacurve.ai — DeepSWE, GPT-6 Astra
- GPT-6 Astra59.9 %HLE-Diamond · 2026-09-22 · accuracysource as of 22/09/2026 · benchmark maintainer · model alone · web: no · reasoning: highSource: lastexam.ai — HLE-Diamond, GPT-6 Astra
- GPT-6 Astra65.8 %HLE-Diamond · 2026-09-22 · accuracysource as of 22/09/2026 · benchmark maintainer · model alone · web: no · reasoning: maxSource: lastexam.ai — HLE-Diamond, GPT-6 Astra
Show 9 more
- GPT-6 Astra74.2 %HLE-Diamond · 2026-09-22 · accuracy · reasoning halfsource as of 22/09/2026 · benchmark maintainer · model alone · subset: reasoning partition (500 questions) · web: no · reasoning: highSource: lastexam.ai — HLE-Diamond, GPT-6 Astra
- GPT-6 Astra82.9 %HLE-Diamond · 2026-09-22 · accuracy · toolssource as of 22/09/2026 · benchmark maintainer · model + harness · Codex CLI 0.155.1 · tools: yes · web: yes · reasoning: highSource: lastexam.ai — HLE-Diamond, GPT-6 Astra
- GPT-6 Astra54.8 %Humanity's Last Exam · 2025-04 · accuracyevaluated 09/09/2026 · benchmark maintainer · mode not stated · judge: o3-mini-2025-01-31 · temperature: 0.0 when configurable · reported as „GPT 6 Astra“Source: labs.scale.com — Humanity's Last Exam, GPT-6 Astra
- GPT-6 Astra83.6 %SimpleBench · avg5 · accuracy (average of 5 runs)evaluated 07/09/2026 · benchmark maintainer · model alone · pass@k: AVG@5 · temperature: 0.7 (top-p 0.95)Machine-readable source: simple-bench.com · JavaScript file — SimpleBench, GPT-6 Astra
- GPT-6 Astra91.5 %BrowseComp · v1 · pass@1 accuracysource as of 03/09/2026 · self-reported by the provider · mode not stated · tools: yes · web: yes · pass@k: 1 · reasoning: maximum over effortsSource: web.archive.org — BrowseComp, GPT-6 Astra
- GPT-6 Astra72.6 %OSWorld · 2.0-v2026.08.08 · partial scoresource as of 03/09/2026 · self-reported by the provider · mode not stated · tools: yes · subset: offline set · reasoning: maximum over efforts (chart: max) · computer use: yesSource: web.archive.org — OSWorld, GPT-6 Astra
- GPT-6 Astra97.6 %FrontierMath Tier 4 · v2 · accuracy (mean score)evaluated 30/08/2026 · benchmark maintainer · mode not stated · tools: yes · budget: 1,000,000 tokens hard limit (forced submission at 660,000) · setting: Epoch AI FrontierMath implementation (Inspect): python tool + submit_answer tool · web: no · reasoning: max · reported as „gpt-6-astra_max“Machine-readable source: epoch.ai · ZIP download — FrontierMath Tier 4, GPT-6 Astra
- GPT-6 Astra93.68 %FrontierMath Tiers 1–3 · v2 · accuracy (mean score)evaluated 30/08/2026 · benchmark maintainer · mode not stated · tools: yes · budget: 1,000,000 tokens hard limit (forced submission at 660,000) · setting: Epoch AI FrontierMath implementation (Inspect): python tool + submit_answer tool · web: no · reasoning: max · reported as „gpt-6-astra_max“Machine-readable source: epoch.ai · ZIP download — FrontierMath Tiers 1–3, GPT-6 Astra
- GPT-6 Astra2.94 %FrontierMath: Erdős problems · 2026-08 · accuracy (mean score)evaluated 28/08/2026 · benchmark maintainer · model + harness · Epoch deepagent (Inspect) with bash, text editor, budget tool; Lean 4 + Mathlib, SageMath, Python; offline arXiv snapsho · tools: yes · budget: $300 spend and 72 hours working time per problem · attempts: 1 · web: no · reasoning: max · reported as „gpt-6-astra_max“Machine-readable source: epoch.ai · ZIP download — FrontierMath: Erdős problems, GPT-6 Astra
Where can it be used?
Only recorded availability statements are shown. A missing plan or surface means unknown, not unavailable. Where it applies only to a plan or a region, the row says so. “Rolling out” is not “available”. “Rollout reported” means a rollout was reported, but nothing has supported it for more than 90 days.
| Product | Status | since | Checked | Sources |
|---|---|---|---|---|
| ChatGPTApp · Business | Limited availability | not documented | checked 4 Oct 2026 | |
| ChatGPTApp · Edu | Limited availability | not documented | checked 11 Oct 2026 | 1learn.chatgpt.com |
| ChatGPTApp · Enterprise | Limited availability | not documented | 1learn.chatgpt.com | |
| ChatGPTApp · Pro 100 | Limited availability | not documented | checked 4 Oct 2026 | |
| ChatGPTApp · Pro 200 | Limited availability | not documented | checked 4 Oct 2026 | |
| ChatGPT WorkWork · Edu | Limited availability | not documented | checked 11 Oct 2026 | 1learn.chatgpt.com |
| ChatGPT WorkWork · Enterprise | Limited availability | not documented | checked 11 Oct 2026 | 1learn.chatgpt.com |
| ChatGPT WorkWork · Plus | Limited availability | not documented | checked 4 Oct 2026 | |
| ChatGPT WorkWork · Pro | Limited availability | not documented | checked 4 Oct 2026 | |
| CodexCodex · Edu | Limited availability | not documented | checked 11 Oct 2026 | 1learn.chatgpt.com |
| CodexCodex · Enterprise | Limited availability | not documented | checked 11 Oct 2026 | 1learn.chatgpt.com |
| CodexCodex · Plus | Limited availability | not documented | checked 4 Oct 2026 | |
| CodexCodex · Pro | Limited availability | not documented | checked 4 Oct 2026 | |
| GitHub CopilotCopilot · Business | Limited availability | not documented | checked 11 Oct 2026 | 1docs.github.com |
| GitHub CopilotCopilot · Enterprise | Limited availability | not documented | checked 11 Oct 2026 | 1docs.github.com |
| GitHub CopilotCopilot · Max | Limited availability | not documented | checked 11 Oct 2026 | 1docs.github.com |
| GitHub CopilotCopilot · Pro Plus | Limited availability | not documented | checked 11 Oct 2026 | 1docs.github.com |
| GitHub Copilot | Limited availability | not documented | checked 11 Oct 2026 | 1docs.github.com |
| Microsoft 365 CopilotCowork | Limited availability | not documented | checked 11 Oct 2026 | 1learn.microsoft.com |
Where a start is “not documented”, a source check found the state in place; the source names no start date.
No statement on record for 9 further products: Claude, Claude Code, Claude Enterprise, Claude Team, Gemini, Gemini for Google Workspace, Grok, Langdock, Le Chat Enterprise. No statement is not the same as not available.
Developer APIs
Developer access — not a statement about any work product.
| Product | Status | since | Checked | Sources |
|---|---|---|---|---|
| OpenAI APIAPI | Available | not documented | checked 3 Oct 2026 | 1developers.openai.com |
What did we test?
No lab run of GPT-6 Astra is published yet.
Nothing is shown until a real run exists. See what the lab tests:
Lifecycle and history
The dates the provider confirmed, and every recorded period of availability. A current period shows its checked status; one without a source check is marked not verified.
Announced
Released
Deprecated—not recorded
Retired—not recorded
What changed?
Sourced changes concerning this model, newest first.
- 9 days ago
- 9 days ago
- 42 days agoadded to the record on 29 Sept 2026
1developers.openai.com
- 42 days agoadded to the record on 29 Sept 2026
1developers.openai.com
Evidence
Every source document behind the facts on this page, by domain.
| Domain | Docs | Documents |
|---|---|---|
| developers.openai.com | 2 | |
| docs.github.com | 1 | |
| help.openai.com | 1 | |
| learn.chatgpt.com | 1 | |
| learn.microsoft.com | 1 |