GPT-6.2 Sol12.389 NLL
GPT-6.2 system cardaggregation: median · lower is better: true
OpenAI ↗Predicts future amino-acid mutations in an engineered phage–plasmid co-evolution experiment.
| Model / creator | Version | Reported score |
|---|---|---|
| GPT-6.2 system card | 11.645 NLL | |
| GPT-6.2 system card | 12.389 NLL |
aggregation: median · lower is better: true
OpenAI ↗aggregation: median · lower is better: true
OpenAI ↗Tests whether a customer-service agent can resolve requests while using tools and following company policies, helping you assess opportunities to reduce manual support work.
Terminal-Bench-ScienceScientific work in a terminal.
LiveBench: DesignTests document and presentation creation, helping you choose a model for preparing client-facing materials with less manual production work.