--- format: "aidr-story-markdown/v1" id: "0a2f1e698b835996756a972d8c79c3f96bd415feb1d109baedc3aa37089ebe72" canonical_url: "https://aidr.today/0a2f1e69?lang=en" title: "OpenAI Plans Broad GPT-6 Astra Access After Messy Rollout" lang: "en" requested_lang: "en" available_langs: ["en","vi"] translation_fallback: null fallback_fields: [] published_at: "2026-09-04T08:47:15.000Z" category: "Models" topics: ["openai","gpt","agent","coding","benchmark","copilot","safety"] source_urls: ["https://x.com/sama/status/2095678759651438887","https://x.com/stevibe/status/2095685429706281424","https://x.com/OpenAIDevs/status/2095676629322494115","https://x.com/sharifshameem/status/2095688352075075878","https://x.com/ValsAI/status/2095732151300088142","https://x.com/TheHackersNews/status/2095766365856731310","https://x.com/sama/status/2095680763039740185","https://x.com/_NathanCalvin/status/2095711871613505823"] summary: "Paid ChatGPT users will get a banked reset for every day Astra remains unavailable on their plan, starting today, in OpenAI's response to complaints about the model's launch. Sam Altman said the company should begin opening GPT-6 Astra to API customers and ChatGPT subscribers in the near future, starting with Pro users. OpenAI has presented Astra as a fast computer-using model, and developers said new Responses API features including async function calling are releasing with it. In one shared demo, Astra researched hundreds of photos of San Francisco's Palace of Fine Arts and iterated on a Blender scene overnight. Vals said Astra led its Code Migration benchmark at 68% accuracy, 10 points above second place, and ran 2-4x faster than comparable models. OpenAI also said it is committing $1 billion to subsidize Daybreak access and frontier capabilities. Astra scored 100% on ExploitBench, and OpenAI says separate exploit-development tests included two zero-day vulnerabilities, though the released version is limited to secure code review and patching and refuses requests to create PoC exploits. A safety post shared by an OpenAI researcher said GPT-6 is significantly better aligned tha" --- # OpenAI Plans Broad GPT\-6 Astra Access After Messy Rollout > [Open the canonical story]() **Published:** 2026-09-04T08:47:15.000Z **Category:** Models **Topics:** openai, gpt, agent, coding, benchmark, copilot, safety ## Summary Paid ChatGPT users will get a banked reset for every day Astra remains unavailable on their plan, starting today, in OpenAI's response to complaints about the model's launch\. Sam Altman said the company should begin opening GPT\-6 Astra to API customers and ChatGPT subscribers in the near future, starting with Pro users\. OpenAI has presented Astra as a fast computer\-using model, and developers said new Responses API features including async function calling are releasing with it\. In one shared demo, Astra researched hundreds of photos of San Francisco's Palace of Fine Arts and iterated on a Blender scene overnight\. Vals said Astra led its Code Migration benchmark at 68% accuracy, 10 points above second place, and ran 2\-4x faster than comparable models\. OpenAI also said it is committing $1 billion to subsidize Daybreak access and frontier capabilities\. Astra scored 100% on ExploitBench, and OpenAI says separate exploit\-development tests included two zero\-day vulnerabilities, though the released version is limited to secure code review and patching and refuses requests to create PoC exploits\. A safety post shared by an OpenAI researcher said GPT\-6 is significantly better aligned tha ## Sources - [Story source]() - [Supporting source]() - [Story source]() - [Supporting source]() - [Supporting source]() - [Supporting source]() - [Story source]() - [Supporting source]()