LiveBench — Instruction Following
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.
| Model | LiveBench — Instruction Following | Observed | Source |
|---|---|---|---|
| Gemini 3.8 FlashGoogle | 81.4 | 2026-09-07 | source ↗ |
| Gemini 3.1 Pro PreviewGoogle | 79.1 | 2026-09-07 | source ↗ |
| Claude Fable 5Anthropic | 75.8 | 2026-09-07 | source ↗ |
| GPT-6 AstraOpenAI | 75.6 | 2026-09-07 | source ↗ |
| Qwen3.8 Max (0902)Alibaba Qwen | 74.1 | 2026-09-07 | source ↗ |
| Claude Fable 5.1Anthropic | 73.0 | 2026-09-07 | source ↗ |
| Claude Opus 4.8Anthropic | 72.0 | 2026-09-07 | source ↗ |
| Grok 4.6SpaceXAI | 71.9 | 2026-09-07 | source ↗ |
| GPT-5.6 SolOpenAI | 71.8 | 2026-09-07 | source ↗ |
| Kimi K3Moonshot AI | 71.4 | 2026-09-07 | source ↗ |
| GPT-5.4OpenAI | 70.2 | 2026-09-07 | source ↗ |
| GLM-5.3Z.ai | 69.3 | 2026-09-07 | source ↗ |
| DeepSeek V4 Pro (0813)DeepSeek | 67.7 | 2026-09-07 | source ↗ |
| Claude Opus 4.7Anthropic | 66.7 | 2026-09-07 | source ↗ |
| GPT-5.6 TerraOpenAI | 64.6 | 2026-09-07 | source ↗ |
| Claude Sonnet 5Anthropic | 63.9 | 2026-09-07 | source ↗ |
| Claude Opus 5Anthropic | 63.8 | 2026-09-07 | source ↗ |
| GPT-5.6 LunaOpenAI | 60.1 | 2026-09-07 | source ↗ |
— = 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.