OpenAI launches GPT-6 Astra

OpenAI released GPT-6 Astra on 3 September 2026. The model is state-of-the-art on computer use, browsing, software engineering, cybersecurity, science and professional work, according to the company.

Astra saturates FrontierMath Tier 4 at 98%, ARC-AGI-3 at 99.9% and ExploitBench at 100%. It reaches 64.6% on Terminal-Bench Science 0.1 against 52.6% for Claude Fable 5.1, at approximately 31% lower estimated API cost, and 59.3% on Agents' Last Exam against 55.5% for Claude Opus 5.

Batch and Flex halve standard rates; Fast mode doubles them at up to 2x speed. Prompts beyond 272K input tokens reprice the entire request. Context is 1,050,000 tokens with 128K maximum output and an April 2026 knowledge cutoff.

OpenAI built a new evaluation informed by the Hugging Face incident, testing whether a model facing a difficult or impossible task goes beyond its intended scope.

OpenAI says Astra meets the 'Critical' cybersecurity threshold under its Preparedness Framework. The public model refuses advanced cyber tasks including proof-of-concept exploits; looser safeguards are reserved for vetted organisations via the Daybreak programme.

The headline 99.9% ARC-AGI-3 score came from a provider-specific harness preserving reasoning state between actions. A neutral harness produced 62.7% at far higher compute cost.

Greg Kamradt of the ARC Prize Foundation said Astra surpassed the foundation's human action-efficiency baseline on 96% of levels, effectively reaching human parity on the benchmark, and described it as a step change in frontier-model performance.

At the launch briefing OpenAI president Greg Brockman argued buyers should judge price per completed task rather than per token, citing OSWorld 2.0 results where Astra scores higher than GPT-5.6 Sol and finishes faster.

Astra launched in the same week as Anthropic's Claude Fable 5.1 and Mythos 5.1 and updates from Meta and Google, prompting CNBC coverage of 'model fatigue' among enterprise buyers on 6 September.

Claude Fable 5.1 matches Astra's $10/$50 headline pricing but charges $0.25 per million cached input tokens against Astra's $1.