OpenAI's changelog, access guide, pricing page and system card confirm release, access, pricing, weight provenance and safety-evaluation boundaries.
OpenAI releases GPT-6 Rosalind Research under trusted access
2026-10-30Verified announcementWATCH
OpenAI released GPT-6 Rosalind Research for approved organizations and authorized users in Work, Codex and the API. It is not available in Chat, and Enterprise deployments begin with a two-week Admin Preview. The model uses GPT-6 Astra weights without additional training or post-training.
Official · Documentationdevelopers.openai.com · changelogOfficial · DocumentationGPT-6 Rosalind Research model
Availability: Limited trusted access for approved Research organizations and authorized users in Work, Codex and the API; not in Chat
What this means for you
For life-science teams, this makes a higher-capability biological reasoning configuration callable, but access review, governance and human oversight are part of the product boundary. Publisher safety evaluations do not establish independent research outcomes.
Evidence & related material
Technical details & scores
| Availability | Limited trusted access for approved Research organizations and authorized users in Work, Codex and the API; not in Chat |
|---|---|
| Enterprise Stage | Two-week Admin Preview; off by default |
| Pricing Per Million Tokens | input_usd 15 cached_input_usd 1.5 output_usd 75 |
| Model Basis | GPT-6 Astra weights; no additional training or post-training |
| Safety Stage | OpenAI classifies Biological/Chemical capability as High; no Critical biological threshold was met in four publisher evaluations |
| Time Basis | Official sources expose only 2026-10-30. The changelog entry was absent from the 2026-10-30T16:01:56Z capture and present in this run, so it is a late discovery without hour-level publication precision. |
Technical impact 4
Engineering value 3
Adoption signal 2
Maturity 3
Verification cost 4
Risk 5