GPT-6 Astra
OpenAI flagship ("most capable model, built for the hardest end-to-end work"). API changelog: Sep 3 2026 (limited preview Sep 3, public Sep 4). Single snapshot gpt-6-astra. OpenRouter also lists openai/gpt-6-astra-pro = same model with reasoning.mode pro. Endpoints: Chat Completions, Responses, Batch.
- Context window
- 1,050,000 tokens
- Max output
- 128,000 tokens
- Knowledge cutoff
- 2026-04
- Input
- text, image
- Output
- text
- License
- proprietary
- Pricing
- input: $10 · cached input: $1 · output: $50 · cache write: $12.5 · unit note: prompts >272K input tokens billed at 2x input and cache rates (per 1M tokens (USD), standard tier) source
- Verified
- 2026-09-29
How to call it
| Provider | Model id | Endpoint / URL | Docs |
|---|---|---|---|
| OpenAI API | gpt-6-astra | https://api.openai.com/v1/responses | docs |
| Azure OpenAI (Microsoft Foundry) | gpt-6-astra | — | docs |
| OpenRouter | openai/gpt-6-astra | openrouter.ai/openai/gpt-6-astra | — |
| Web app | — | chatgpt.com | — |
Notable capabilities (6)
- Max reasoning effort: reasoning.effort adds a new "max" level above xhigh (low/medium/high/xhigh/max). source
- 1M-token context: 1.05M context window (922K max input) with 128K output on the flagship. source
- Restricted cyber behaviour: Released as a restricted version that rejects certain cybersecurity prompts; separate Cyber/Daybreak models exist for that domain. source
- Recurrent-depth reasoning (found after launch): Reported new "recurrent depth" technique that obscures some of the reasoning, raising monitorability concerns among safety researchers. source
- Ultrafast service tier (found after launch): From Sept 29 2026, service_tier "ultrafast" gives up to 6x faster generation in the API (8x / ~300 tok/s in Codex) at 6x price: $60 input / $6 cached / $75 cache write / $300 output per 1M tokens (<=272K ctx); default limits 500K-5M TPM. source
- Powers dots always-on agents (found after launch): OpenAI's dots (launched Sept 29 2026) run on GPT-6 Astra, each with its own cloud computer; also the default model in Agents API computer-use examples. source
Hardest end-to-end work: long-horizon agentic coding, research, analysis, computer use.
curl https://api.openai.com/v1/responses \
-H "Authorization: Bearer $OPENAI_API_KEY" -H "Content-Type: application/json" \
-d '{"model":"gpt-6-astra","input":"Hello"}'
Sources:
- https://developers.openai.com/api/docs/models/gpt-6-astra
- https://developers.openai.com/api/docs/pricing
- https://developers.openai.com/api/docs/changelog
- https://en.wikipedia.org/wiki/GPT-6_Astra
- Ultrafast: https://developers.openai.com/api/docs/guides/ultrafast-mode
- Models overview: https://developers.openai.com/api/docs/models
Other OpenAI models
GPT-6.1 Sol · GPT-6 Luna · GPT-6 Sol · GPT Image 2.5 Flare · GPT Image 2.5 Sunburst · GPT-Live-Transcribe · GPT-Transcribe · GPT-5.6 Terra · GPT-Live 1 · GPT-Realtime-2.1 · GPT-Realtime-2 · GPT-Realtime-Translate · GPT-Rosalind · GPT-Audio-1.5 (and gpt-audio / gpt-audio-mini) · GPT-5.3-Codex · gpt-oss-120b · gpt-oss-20b · GPT-4o mini TTS · text-embedding-3-large · text-embedding-3-small · GPT-5.6 Luna · GPT-5.6 Sol · GPT-Realtime-Whisper · GPT-5.5 Pro · GPT-5.5 · GPT Image 2 · GPT-5.4 · GPT-Realtime-1.5 · GPT-4.1 · GPT-4o · TTS-1 / TTS-1 HD · Whisper large-v3 / large-v3-turbo (open weights) · GPT-Realtime and GPT-Realtime mini · o3 · GPT-4o Transcribe / Mini Transcribe / Transcribe Diarize · Whisper (whisper-1 API) · Sora 2