GPT-6 Astra
OpenAI’s flagship model for difficult end-to-end professional work across reasoning, coding, computer use, browsing, research, cybersecurity, science, and document creation.
- PROVIDER
- OpenAI
- RELEASE DATE
- September 3, 2026
- CONTEXT WINDOW / MAX OUTPUT
- 1.05M context · 128K maximum output
- PRICING
- $10 input · $1 cached input · $12.50 cache write · $50 output per 1M tokens
- PRICING NOTES
- Requests above 272K input tokens: 2× input/cache and 1.5× output. Batch and Flex: 50% of Standard. Fast mode: 2× applicable rates.
- KNOWLEDGE CUTOFF
- April 30, 2026
- MODALITIES
- Text input/output · image input; audio and video not supported
- ARCHITECTURE
- Closed-weight reasoning and agent model · reasoning-token support
- PARAMETERS
- Not published
- ACCESS
- LIMITED ROLLOUT
- BEST FOR
- Complex agents · computer use · coding · research · professional work
- LATEST UPDATE
- OpenAI model documentation and launch materials reviewed September 3, 2026
- CATALOG VERIFIED
- 2026-09-03
Astra is designed to carry multistep work across code, browsers, and professional software with stronger computer-use performance and more deliberate instruction handling. It supports the full Responses API tool stack, but tool calling requires Responses rather than Chat Completions.
Sources
Related
- GPT-5.6 Sol OpenAI