AI Agent StoreOpenAI · Restricted research access
GPT-6 Rosalind Research
Help approved life-sciences teams analyze evidence, plan experiments and work with scientific tools.
Released 2026-10-30 · First reviewed 2026-10-31 · Checked 2026-11-03.
We still need comparable independent tests. You can compare the confirmed features and limits below, but we cannot yet give this version an Agent Fit or Practical Value rating. Test it on your own tasks before switching; results for another version or thinking setting may not apply.
What is announced so far
Token prices measure the small pieces of text the model reads or produces. Capacity is the amount the service accepts in a request; check the limits below for any fallback to another model. A dash means the rate is unverified.
Exact provider API ID: gpt-6-rosalind. Specialized biological reasoning and scientific tool use. Public context limits, output limits, input formats and thinking settings are not verified.
Requires Rosalind Research approval for the organization and user. Work and Codex access does not automatically include API access, which requires separate approval. Discovery approval is insufficient. Approved use is internal research; customers cannot expose this access to outside users.
The pricing table distinguishes short and long context, but the reviewed sources do not establish this model’s threshold or capacity. Confirm the limit for your approved account before sending large documents.
Short-context rates per million tokens: $15 input, $1.50 cache reads and $75 output. Long-context rates: $30/$3/$112.50. Cache writes cost $18.75 for 30 minutes or $37.50 for 12 hours at short context; long-context writes cost $37.50/$75. Confirm eligibility and the context boundary before estimating a large research job. Allow for retries and any tool fees. Models that think before answering may also charge for that work.
Provider release and specifications ↗
Check availability and plans ↗
More setup and licensing details ↗
Test it before switching
Give each model the same examples, tools and definition of success. Record correct outcomes, response time, retries and total cost. A cheaper token price does not guarantee a cheaper completed task.
Compare caching and long-context costs ↗