HumanEval.org

Benchmarks

LiveBench — Mathematics

last retrieved 2026-09-08 21:40 UTC

Third-party scores shown as context with provenance — our own arena ratings from blind human votes stay alongside every row and remain the primary signal.

ModelLiveBench — MathematicsObservedSource
Claude Fable 5.1Anthropic97.02026-09-07source ↗
GPT-6 AstraOpenAI96.82026-09-07source ↗
GPT-5.6 SolOpenAI96.22026-09-07source ↗
Claude Fable 5Anthropic96.02026-09-07source ↗
Claude Opus 5Anthropic95.72026-09-07source ↗
DeepSeek V4 Pro (0813)DeepSeek95.12026-09-07source ↗
GPT-5.6 TerraOpenAI94.92026-09-07source ↗
Claude Opus 4.8Anthropic94.32026-09-07source ↗
GPT-5.4OpenAI94.22026-09-07source ↗
Claude Sonnet 5Anthropic92.92026-09-07source ↗
Claude Opus 4.7Anthropic92.82026-09-07source ↗
Grok 4.6SpaceXAI92.62026-09-07source ↗
Gemini 3.8 FlashGoogle91.62026-09-07source ↗
Qwen3.8 Max (0902)Alibaba Qwen91.32026-09-07source ↗
Gemini 3.1 Pro PreviewGoogle91.02026-09-07source ↗
GLM-5.3Z.ai87.92026-09-07source ↗
GPT-5.6 LunaOpenAI87.22026-09-07source ↗
Kimi K3Moonshot AI84.42026-09-07source ↗

— = not yet evaluated by HumanEval (no eligible arena votes). External values are the newest observation per model; hover a value for source notes and a rating for its 95% CI. Unit: score, higher is better.

Not enough overlap with our arena ratings for a correlation scatter yet — fewer than two scored models are rated.