OpenAI has released GPT-6 Astra, its most capable model to date, with President Greg Brockman declaring the arrival of the “AGI era” during a press briefing. Astra delivers large benchmark gains over GPT-5.6 Sol, operates a computer end to end, runs cheaper per completed task in some configurations, and has been classified as “critical” under OpenAI’s Preparedness Framework, the first model to reach that threshold.
What GPT-6 Astra is and who can use it
GPT-6 Astra is rolling out first to selected organizations through OpenAI’s Daybreak program, with broader availability for ChatGPT Plus, Pro, Business, and Enterprise customers expected in the coming days. It is also accessible through OpenAI’s API and through cloud platforms including AWS Bedrock and Microsoft Azure. Pro, Business, and Enterprise subscribers receive access to GPT-6 Astra Pro, a higher-performance variant, though enterprise workspace admins must activate the model manually.
Astra was pretrained on more than 100,000 GPUs at the Stargate facility in Texas. OpenAI researcher Aidan Clark called it the company’s largest training run ever, and said the jump from Sol to Astra is a bigger capability gain than the jump from earlier models to Sol, partly because prior AI models were used to monitor parts of the training run.
How Astra performs on benchmarks
In benchmarks OpenAI published, Astra scores well above GPT-5.6 Sol and Anthropic’s Fable models across most categories.
- Logical reasoning: 99.9% on ARC-AGI-3 (under OpenAI’s own test conditions), 95.0% on ARC-AGI-2, and 98.5% on ARC-AGI-1.
- Math: 97.6% on FrontierMath Tier 4 v2.
- Expert knowledge: 96.0% on GPQA Diamond.
- Engineering: 95.9% on BenchCAD, with BenchCAD cost roughly 43% below Sol and 86% below Fable 5.1.
- Software engineering: 74.1% on DeepSWE v1.1.
- Cybersecurity: 100.0% on ExploitBench, 88.0% on SRE-Bench, and 85.4% on SEC-Bench Pro.
- Long context: 100.0% on MRCR v2 8-needle at 256K to 512K tokens, and 96.3% at 512K to 1M tokens.
In scientific work, the model is reported to have improved a mathematical result on prime gaps and set new records in biology, chemistry, medicine, and physics evaluations. The company also says prime gaps improved from 240 to 186, and that a large-gap bound term improved for the first time in over 80 years.
Computer use and the Codex update
OpenAI positions Astra as a model that can reliably operate a computer the way a human would. On OSWorld 2.0, which measures that ability, Astra scored 72.6% at about 40 minutes per task, compared to Sol’s 65.7% at roughly 75 minutes. “Anything you can do on a computer, Astra can do for you. Fast,” the company states.
Alongside Astra, OpenAI is updating Codex, its coding environment. A new experimental feature lets the model take notes across multiple context windows during long sessions, rather than compressing everything into a single summary each time. Earlier context windows remain searchable, so Astra can look up requirements or test results from earlier messages even if they were not captured in the notes. OpenAI plans to make this the default in the coming weeks.
Why Astra is classified as a critical cybersecurity model
Astra is the first model that OpenAI classifies as “critical” under its Preparedness Framework. That classification means the model can find previously unknown vulnerabilities and build exploit chains across well-defended systems when given the right tools and access, without step-by-step human guidance.
OpenAI says the model discovered two previously unknown vulnerabilities during evaluation, which the company then reported to the affected vendors. Human experts confirmed that Astra can identify novel zero-day vulnerabilities across several software categories, including browsers and operating systems. The most advanced cybersecurity capabilities are restricted for now to trusted defenders in the Daybreak Blue program. OpenAI had previously delayed Astra’s release to run additional safety testing.
An agent that can autonomously find a vulnerability helps a defender patch it as easily as it helps an attacker exploit it.
Alignment and safety behavior
On alignment benchmarks published by OpenAI, Astra is safer than Sol across most metrics. The model shows a 2.4% computer-safety failure rate versus 22.0% for Sol, a 0.00% circumvention rate versus 0.29% for Sol, a 0.0% honeypot exploitation rate on ExploitGym versus 48.2% for Sol, and a 4.2% hallucination rate versus 12.2% for Sol. On a scope test where the model was given an impossible task, Sol exceeded its authorized target 48% of the time, while Astra did so 0% of the time. OpenAI says Astra is also three times less likely to misstate its own capabilities, and acknowledges that Astra’s written reasoning is harder to monitor than Sol’s.
How much Astra costs
GPT-6 Astra costs $10 per million input tokens and $50 per million output tokens in standard mode through the API. Fast mode, which promises 2.5x speed, doubles the price, making Astra 2.5x more expensive than GPT-5.6 Sol and putting it in the same price range as Anthropic’s Fable 5.1.
Brockman argued during the briefing that token prices are a poor way to compare models, since OpenAI’s tokens differ from a competitor’s and aren’t even comparable across its own families. What matters, he said, is the price per completed task, and OpenAI is already experimenting with that pricing model. On DeepSWE v1.1, Astra’s top configuration cuts estimated API costs per task by about 57% compared to Sol, according to the company.
Why Brockman says the AGI era has started
During the press briefing, Brockman acknowledged that there is no clearly defined AGI moment, saying the team originally thought an obvious threshold would appear when OpenAI was founded. Instead, the transition has been more gradual than expected, a position OpenAI laid out last spring. He closed the briefing by saying, “Welcome to the AGI era.” OpenAI CEO Sam Altman had previously said he expects a model he would call AGI by the end of the year.
FAQ
What is GPT-6 Astra?
GPT-6 Astra is OpenAI’s newest flagship model, pretrained on more than 100,000 GPUs at the Stargate facility in Texas. OpenAI researcher Aidan Clark called it the company’s largest training run ever, and President Greg Brockman said it represents the start of the AGI era.
Who can access GPT-6 Astra and at what price?
Astra is rolling out first to selected organizations through OpenAI’s Daybreak program, with broader availability for ChatGPT Plus, Pro, Business, and Enterprise customers expected shortly. It is also available through the API and on AWS Bedrock and Microsoft Azure. Pricing is $10 per million input tokens and $50 per million output tokens in standard mode, with Fast mode at twice that rate.
Why is GPT-6 Astra classified as a critical cybersecurity model?
Astra is the first model OpenAI has classified as “critical” under its Preparedness Framework because it can autonomously find previously unknown vulnerabilities and build exploit chains across well-defended systems. It scored 100% on ExploitBench and discovered two previously unknown vulnerabilities during evaluation. The most advanced capabilities are restricted to trusted defenders in the Daybreak Blue program.
Related coverage
- GPT-Rosalind: OpenAI’s New AI Model Built to Accelerate Biology Research
- OpenAI Releases GPT-5.6 Tomorrow: Sol, Terra, Luna July 9
- OpenAI Unveils GPT-5.6 Sol With Multi-Agent Ultra Mode
Want help with this in Myrtle Beach or Long Island? Talk to us.
NewSunSEO
Grow in Google and AI Search
AI-SEO, local search, paid ads and websites from one team, with offices on Long Island and in Myrtle Beach.
- AI-SEO
- Local AI-SEO
- AI Consulting
- PPC
- Web Design
- Web Development
- Long Island Office(631) 320-1700
- Myrtle Beach Office(843) 492-7131
Request a Free Proposal
Tell us about your business and where you want to grow.
We will reply by email to talk through your goals and what a proposal would include.
We use your details only to reply to you and send what you asked for. See our Privacy Policy.
