GPT-6 Astra Expands to Paid Users After September Launch

OpenAI released GPT-6 Astra on September 3 to Daybreak users before expanding to paid plans the next day.

On September 4, 2026, GPT-6 Astra became available to all Pro, Enterprise, and Business Premium users. The Verge reported that Sam Altman apologized for a “messy” rollout that left paying users locked out.

The rollout began with a September 3 system card on the Deployment Safety Hub and reached standard paid plans within 24 hours.

The sequence began on September 3 when OpenAI released the model and published its system card on the Deployment Safety Hub. TechCrunch noted that the launch made the model available Thursday to customers that use Daybreak, its cybersecurity program. Over the next week it will also become available through OpenAI���s paid plans including Pro, Plus, Enterprise, and Business accounts as well as through its API. According to developers.openai.com, GPT-6 Astra is the most capable model built for the hardest end-to-end work including complex reasoning, coding, computer use, research, and document creation. Greg Brockman stated during a journalist call that Astra is the company’s “most intelligent and, also very importantly, our most aligned model yet” and that it “brings together years of our research and big bets, with each breakthrough having built on the last” with unmatched speed, accuracy, and safety.

According to developers.openai.com, the model ID is gpt-6-astra with a 1,050,000 context window, maximum 922,000 input tokens, and 128,000 output tokens. Input modalities cover text and image while output remains text only. The knowledge cutoff sits at April 30, 2026. Supported features include streaming, structured outputs, function calling, file search, image input, web search, and prompt caching. Tools available through the Responses API encompass web search, file search, image generation, code interpreter, hosted shell, apply patch, skills, computer use, mcp, and tool search. Endpoints support Chat Completions, Responses, and Batch but exclude Realtime, Realtime translation, Realtime transcription, Assistants, Fine-tuning, Embeddings, image generation, videos, image edit, speech generation, transcription, translation, and moderation.

Prompts with more than 272K input tokens are priced at 2x input and cache rates and 1.5x output for the full request. Cache writes are billed at 1.25x the uncached input token rate. Batch and Flex are priced at 50% of Standard rates. Fast mode is priced at 2x the applicable rates.

TechCrunch coverage highlighted that Astra scores higher than other existing models including OpenAI’s own Sol and Anthropic’s Fable when it comes to activities like finding bugs, executing terminal tasks, and answering queries about codebases. The model also incorporates opaque recurrence, which reduces chain-of-thought visibility and raises monitoring challenges according to chief scientist Jakub Pachocki. Brockman framed the technique as a natural result of capability growth where harder tasks require fewer language tokens. The Verge post confirmed availability reached all listed paid plans by the evening of September 4. Continued rollout now depends on sustained performance within the safety evaluations published on deploymentsafety.openai.com.

The September 3 system card release alongside the Daybreak debut established safety documentation as a prerequisite for each expansion step.

What's underappreciated here: the September 3 system card release alongside the Daybreak debut established safety documentation as a prerequisite for each expansion step rather than an afterthought. This sequence aligns with the pattern of controlled deployment that preceded GPT-5.6 family releases earlier in 2026. TechCrunch recorded Brockman's comment that AGI had shifted from a contractual trigger in the Microsoft partnership to a mission concept, removing any automatic dissolution clause. The model therefore ships under existing enterprise agreements without new legal thresholds. The access pattern points to API availability for remaining paid tiers in the coming week. Continued rollout now depends on sustained performance within the safety evaluations published on deploymentsafety.openai.com.

A reversal that limited Daybreak organizations to the original September 3 cohort would contradict the documented expansion.

The sequence of events leaves open whether further rate-limit increases or additional endpoints will follow the same 24-hour cadence once the remaining paid tiers receive access. The pricing page lists no figure for the enterprise tier beyond the standard token rates already shown. No source discloses exact batch queue behavior for Tier 5 users after the first week of access.