⌕Is ___ SOTA?
benchmarks / terminal-bench-2

Terminal-Bench 2.0

accuracy. Status display_only. Contamination risk med. Maintainer Terminal-Bench / Epoch collection. Source.

Display only.This benchmark cannot decide a crown under methodology v1.0.
ModelScoreDateKind
GPT-5.50.8471910112362026-04-23observation
GPT-5.40.8182026-03-12observation
Gemini 3.1 Pro Preview0.7842026-03-02observation
Claude Opus 4.70.80224719101100012026-04-16observation
Claude Opus 4.60.7982026-03-12observation
GPT-5.3 Codex0.78426966292100012026-02-05observation
Gemini 3 Pro Preview0.69425717852700012025-11-18observation
GPT-5.2 Codex0.66516853932600012026-01-14observation
GPT-5.20.64943820224700012025-12-11observation
GPT-5.20.6492025-12-24observation
Gemini 3 Flash Preview0.512025-12-23observation
Claude Opus 4.50.6312025-12-11observation
GPT-5.1 Codex Mini0.4312025-11-17observation
GPT-5.1 Codex0.6044943820222025-11-19observation
Claude Opus 4.5 20251101 128k0.5912025-12-17observation
GPT-5.1 Codex0.3692025-11-17observation
Grok 4.200.5732026-04-02observation
Claude Sonnet 4.60.5342696629212026-02-17observation
GLM 50.5241573033712026-02-11observation
GPT-50.496067415732025-08-07observation
GPT-50.4962025-11-04observation
GPT-5.10.4762025-11-16observation
GPT-5.10.4758426966292025-11-12observation
Claude Sonnet 4.50.4012025-11-04observation
Minimax M2 70.450561797753000042026-03-18observation
GPT-5 Codex0.4432025-11-04observation
Kimi K2 50.432022471912026-01-27observation
Minimax M2 50.426966292135000042026-02-12observation
Deepseek/deepseek V3.20.3955056179782025-12-01observation
Claude Opus 4.10.3482025-11-04observation