A September 5, 2026 Medium analysis by Mehul Gupta recaps OpenAI's GPT-6 Astra launch and the headline benchmark figures that dominated coverage: 99.9% on ARC-AGI-3, 97.6% on FrontierMath Tier 4, and 100% on ExploitBench. The piece notes that of the fourteen benchmarks OpenAI highlighted in its announcement, only one o The analysis argues for looking past composite scores to implementation details when evaluating model progress, positioning itself as a case study rather than a takedown of Astra. It acknowledges that GPT-6 Astra represents a genuine step forward in some respects, while encouraging developers and evaluators to scrutini

A Vellum breakdown of OpenAI's published GPT-6 Astra benchmark tables details that the model was trained on OpenAI's largest run to date, exceeding 100,000 GPUs at the Stargate site in Texas, and is the first release where earlier OpenAI models supervised training of the new one. Computer-use scores include 59.3 on Age The Vellum analysis notes that Meta's Muse Spark 1.3 edges Astra on agentic coding and that the Artificial Analysis index places Claude ahead, while OpenAI characterizes Astra as "the world's most intelligent and aligned model" and President Greg Brockman told reporters it is "not unreasonable to feel that we are now i

OpenAI on September 3, 2026 announced the rollout of GPT-6 Astra, its latest model that CEO Sam Altman described as a "new capability level." The rollout is happening in phases, with companies in OpenAI's application-based cybersecurity program Daybreak receiving first access, followed by availability through ChatGPT P The CNBC report highlights that GPT-6 Astra is OpenAI's first model to reach the company's internal "Critical" cybersecurity threshold, a designation that has prompted restricted access to its advanced cyber capabilities. OpenAI President Greg Brockman stated the company is investing heavily in safety and alignment, an

Artificial Analysis published a September 3, 2026 benchmarking study finding that GPT-6 Astra makes significant gains in the Coding Agent Index, scoring 67 (approximately equal to Claude Opus 5 and Fable 5 in Claude Code, and Muse Spark 1.3 in Muse Code), while Fable 5.1 in Claude Code leads at 70. The model is roughly On the Intelligence Index, GPT-6 Astra scores 61, matching GPT-5.6 Sol but trailing Fable 5.1 and Meta's Muse Spark 1.3. API pricing increased 2.5x to $10 per million input tokens and $50 per million output tokens (up from $4/$20), with the same 90% cache-read discount and 25% cache-write premium; at max effort this ma

A Coursiv explainer confirms GPT-6 Astra's official release on September 3, 2026, and lists headline specifications: a 1.05-million-token context window, up to 128,000 output tokens, a reported 99.9% score on ARC-AGI-3, and stronger results across computer-use, mathematics, coding, scientific, and cybersecurity evaluat For developers, the Coursiv summary cites GPT-6 Astra API pricing starting at $10 per million input tokens and $50 per million output tokens, and notes that broader ChatGPT Free tier access has not been announced while rollout continues through paid ChatGPT tiers and the OpenAI API. The piece positions Astra as a subst