OpenAI releases GPT-Rosalind-6 system card addendum
Original title: GPT-Rosalind-6 System Card Addendum
GPT-Rosalind-6 improves biological data analysis with 37.7% single-attempt success on GeneBench Pro v13.
OpenAI released GPT-Rosalind-6, using same weights as GPT-6 Astra with no additional training. It excels in genomics and quantitative biology benchmarks. The model is for trusted-access deployment with enhanced biological safeguards.
- Single-attempt success 37.7% vs 16.7% for GPT-Rosalind-5.5 on GeneBench Pro v13
- Focus on biological and chemical capabilities
- Trusted-access deployment for advanced life sciences
- Monitors for biosecurity benchmarks
- Vendors
- OpenAI
- AI models
- GPT-Rosalind-6
From the original article
3 min read at deploymentsafety.openai.com
GPT-Rosalind-6 uses the same weights as GPT-6 Astra. We did not add training data or perform additional post-training for this release. Details of the underlying model’s data and training are provided in the
By gmays · deploymentsafety.openai.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.