Open-weight translation model from Bilibili's Index team, covering 150 languages and following glossary and style instructions
GPT-6 Astra
New· 4OpenAI's flagship model, released 3 September 2026 with a 1,050,000-token context window and vetted enterprise access first
About GPT-6 Astra
GPT-6 Astra reached users on 3 September 2026 as OpenAI's new flagship, priced as a premium tier above GPT-5.6 Sol, with extra options for cached input, batch jobs and a faster mode at a higher rate.
Capacity and cutoff
The model carries a context window of 1,050,000 tokens and can produce up to 128K tokens of output in a single reply. Its knowledge cutoff is April 2026.
Reported results
On OSWorld 2.0, a computer-use test, OpenAI puts the score at 72.6%, and on FrontierMath Tier 4 at 97.6%. ExploitBench comes out at 100%. ARC-AGI-3 gives two figures rather than one. OpenAI's own Provider Adapter harness yields 99.9%, while the shared Standard harness gives 62.7%, per ARC Prize's report. The same report cautions that a saturated benchmark would not by itself show that AGI had been reached.
Cyber threshold and staged rollout
OpenAI has placed Astra at the critical cybersecurity level of its preparedness framework, the first of its models to be rated that high. The version released to the public declines to write working exploit demonstrations. Rollout began with vetted enterprise customers in a gated programme called Daybreak, ahead of ChatGPT plans and the API.
TrustList summary
GPT-6 Astra is OpenAI's flagship AI model, shipped on 3 September 2026. Its public version refuses requests for proof-of-concept exploits, and its rollout started with vetted enterprise customers in the gated Daybreak programme.
What GPT-6 Astra offers
- Long context: a window of 1,050,000 tokens and up to 128K tokens of output.
- Computer use: OpenAI reports 72.6% on OSWorld 2.0.
- Mathematics: OpenAI reports 97.6% on FrontierMath Tier 4.
- ARC-AGI-3: 99.9% in OpenAI's Provider Adapter setup and 62.7% in the shared standard setup, according to ARC Prize's report. ARC Prize cautions that a high score would not prove AGI.
- Cybersecurity threshold: the first model OpenAI has labelled as reaching its 'critical' cybersecurity level.
Who it works with
Vetted enterprise customers first, followed by ChatGPT plans and the API.
Key facts
- Released: 3 September 2026
- Knowledge cutoff: April 2026
- Developer: OpenAI
Summary written by TrustList from the business's own published information, Nov 2026.
Benchmarks & AI stats
| Benchmark | Official | Community avg |
|---|---|---|
| ARC-AGI-3 (Provider Adapter harness)reasoning | 99.9 % | - |
| ARC-AGI-3 (Standard harness)reasoning | 62.7 % | - |
| Artificial Analysis Intelligence Index v5.0.1llm | 57 index | - |
| ExploitBenchreasoning | 100 % | - |
| FrontierMath Tier 4reasoning | 97.6 % | - |
| OSWorld 2.0agent | 72.6 % | - |
“Official” values are editor-approved and feed the ranking. “Community avg” is the mean of member submissions (shown for transparency; it never affects the ranking until an editor approves a value).
Sign in to add a benchmark score for this model.
GPT-6 Astra launches
Trust Score
An earned signal from verification, reviews, awards, transparency and engagement - the vendor can't buy it.
Updated 9/16/2026
What the business tells TrustList
- Founded
- 2026 (0 years) evidence sent
Request a demo or quote from GPT-6 Astra
Reviews
Write the first review of GPT-6 Astra
Used it? Your experience helps other buyers decide.
Write a reviewQuestions & answers
No questions yet. Be the first to ask about GPT-6 Astra.
GPT-6 Astra alternatives
Similar listings buyers compare against this one.
Open-source 2B decision model from AWS's Strands Agents project that picks options and returns scores with a reliability figure
Anthropic's smallest and fastest Claude 5.5 model, available on its own API and three cloud platforms
Cantina Security's first open-weight security model, a post-trained GLM-5.3-Flash for vulnerability research and exploit checks
Perplexity's open-weight decision model that picks among given options for routing and checks in AI systems
Microsoft AI's text-to-speech model for keeping one voice across 23 languages, with voice cloning from short audio
Google's open multimodal embedding model that maps text, images, audio and video into one vector space, under Apache 2.0
Liquid AI's open-weight decision model, giving yes or no, choice or score answers from text and images
More on TrustList
Everything here links back to the same verified catalogue. Pick your next stop.
- Write a reviewHelp other buyers decide
- CompaniesAgencies, consultancies and IT service providers, ranked by verified reviews.
- ProductsSoftware and SaaS with pricing, features, integrations and alternatives.
- ArticlesGuides, comparisons and research from the editorial desk and the community.
- AwardsAnnual recognition decided by verified reviews and an independent jury.
- LaunchesNew products and releases, voted up by the community every day.
- RequestsBuyers describe what they need; vendors respond directly.
- PeopleReviewers, authors and makers with public profiles.
- ComparePut up to four listings side by side before you shortlist.