openai ships gpt-6 astra, then cheaper gpt-6 sol and luna

what happened

• OpenAI's API changelog records the release of GPT-6 Astra on September 3, 2026, describing it as 'our most capable model, built for the hardest end-to-end work.'[1]
• The gpt-6-astra model page lists a 1,050,000-token context window, 128,000 max output tokens and an April 30, 2026 knowledge cutoff.[2]
• Standard API pricing is $10 per million input tokens, $1 cached input and $50 output; prompts over 272K input tokens cost more, and Batch and Flex are 50% of standard rates.[2,3]
• Per the changelog, gpt-6-astra does not support the none reasoning effort, custom temperature or top_p, or logprobs.[1]
• OpenAI said Astra would roll out first to a limited set of organizations, then over the following days to ChatGPT Plus, Pro, Business and Enterprise users and through the OpenAI API, Microsoft Azure and AWS Bedrock; Enterprise workspace access was off by default at launch until an admin enabled it.[3]
• OpenAI says Astra meets the 'Critical' cybersecurity threshold in its Preparedness Framework and will refuse more advanced cyber tasks such as creating proof-of-concept exploits; when extra safety checks pause a task, ChatGPT and Codex may ask the user to review it, while in the API the task stops.[3]
• On September 22, 2026 OpenAI added gpt-6-sol ($2 input, $0.20 cached, $10 output per million tokens) and gpt-6-luna ($0.10 / $0.01 / $0.50), and on September 29 gpt-6.1-sol at $2 input, $0.10 cached and $10 output.[1,4]
• OpenAI's deprecations page lists GPT-4o (gpt-4o-2024-08-06, gpt-4o-2024-11-20) and GPT-4.1 (gpt-4.1-2025-04-14) for shutdown on January 14, 2027, with gpt-6-luna as the recommended replacement.[5]