OpenAI announced an addendum for GPT-Rosalind-6, a model designed to support legitimate biological research. It uses the same weights as GPT-6 Astra and includes additional deployment controls for advanced life sciences research.
Key facts
| Fact | Detail | The source says |
|---|---|---|
| Model | GPT-Rosalind-6 | “GPT-Rosalind-6 uses the same weights as GPT-6 Astra.” |
| Training Data | None added | “We did not add training data or perform additional post-training for this release.” |
| Access | Trusted access | “Access is limited to approved customers through our trusted-access deployment structure where organizations must demonstrate they are…” |
What happened
OpenAI has released an addendum for GPT-Rosalind-6, a model designed to support legitimate biological research. The addendum details GPT-Rosalind-6's biological capability assessment, additional safety and alignment evaluations, and deployment safeguards. It uses the same weights as GPT-6 Astra, the flagship reasoning model, and did not add training data or perform additional post-training for this release. Access to GPT-Rosalind-6 is limited to approved customers through a trusted-access deployment structure, where organizations must demonstrate they are conducting legitimate scientific research with public benefit, have strong governance and safety oversight, and control access with enterprise-grade security.
What to weigh
- GPT-Rosalind-6 uses the same weights as GPT-6 Astra.
- Access is limited to approved customers.
Source: OpenAI