GPT-6.1 Terra
Terminal-Bench · task success · 50.3% · higher is better
OpenAI · vendor · 27 Oct 2026
Configuration incomplete · not ranked
Sources: OpenAI
Benchmarks
Compare results only within one benchmark and verified evaluation cohort. Vendor claims and independent measurements stay separately labeled. Overlapping confidence intervals do not establish superiority.
5 results
Timestral-123M
GIFT-Eval · MASE · 0.659 ratio · lower is better
Mistral AI · vendor · 25 Oct 2026
Configuration incomplete · not ranked
Sources: Mistral AI
Timestral-209M
GIFT-Eval · MASE · 0.658 ratio · lower is better
Mistral AI · vendor · 25 Oct 2026
Configuration incomplete · not ranked
Sources: Mistral AI
Timestral-28M
GIFT-Eval · MASE · 0.678 ratio · lower is better
Mistral AI · vendor · 25 Oct 2026
Configuration incomplete · not ranked
Sources: Mistral AI
Timestral-5M
GIFT-Eval · MASE · 0.705 ratio · lower is better
Mistral AI · vendor · 25 Oct 2026
Configuration incomplete · not ranked
Sources: Mistral AI