LiveBench what it measures, how it is scored and who leads it

Name: LiveBench
Creator: LiveBench (Abacus.AI, NYU, NVIDIA and others)
Published: 2024-06-27

Reasoning · LiveBench (Abacus.AI, NYU, NVIDIA and others) · introduced Jun 27, 2024 · Global average

Scores from LiveBench, published Jun 25, 2026

Rank	Model	Global average
1	Claude Fable 5.1Anthropic · as “claude-fable-5-1-max-effort”	83.4%
2	Claude Opus 5.5Anthropic · as “claude-opus-5-5-max-effort”	83.2%
3	Claude Fable 5Anthropic · as “claude-fable-5-max-effort”	83.0%
4	GPT-6 AstraOpenAI · as “gpt-6-astra-max”	82.2%
5	Muse Spark 1.3Meta · as “muse-spark-1.3-xhigh”	81.6%
6	DeepSeek V4.1 FlashDeepSeek · as “deepseek-v4.1-flash-max”	81.1%
7	GPT-5.6 SolOpenAI · as “gpt-5.6-sol-max”	81.0%
8	GPT-5.5OpenAI · as “gpt-5.5-xhigh”	80.2%
9	Claude Opus 5Anthropic · as “claude-opus-5-max-effort”	80.1%
10	Smaug Agentic · as “smaug-agentic”	79.5%
11	GPT-6 SolOpenAI · as “gpt-6-sol-max”	79.3%
12	Kimi K3Moonshot AI · as “kimi-k3”	79.2%
13	Gemini 3.7 FlashGoogle · as “gemini-3.7-flash-high”	78.8%
14	Qwen3.8 Max (0902)Alibaba · as “qwen3.8-max”	78.5%
15	Grok 4.6SpaceXAI · as “grok-4.6”	78.0%
16	GPT-5.4OpenAI · as “gpt-5.4-xhigh”	78.0%
17	Muse Spark 1.2Meta · as “muse-spark-1.2-xhigh”	78.0%
18	GPT-5.6 TerraOpenAI · as “gpt-5.6-terra-max”	77.9%
19	DeepSeek V4 Pro 0813DeepSeek · as “deepseek-v4-pro-0813”	77.4%
19	Smaug Flash · as “smaug-flash”	77.4%
21	Grok 4.7SpaceXAI · as “grok-4.7-xhigh”	77.4%
22	Gemini 3.1 Pro PreviewGoogle · as “gemini-3.1-pro-preview-high”	77.0%
23	Smaug Mini · as “smaug-mini”	76.9%
24	DeepSeek V4 Flash VisionDeepSeek · as “deepseek-v4-flash-vision-exp”	76.8%
25	Claude Opus 4.7Anthropic · as “claude-opus-4-7-xhigh-effort”	76.5%

Task: Answer fresh questions in math, coding, reasoning, data analysis, language and instruction following.
Dataset: Questions drawn from recent sources and refreshed regularly so they cannot be in training data.
Method: Every answer scored automatically against an objective ground truth, no model judge.
Metric: Global average
Organization: LiveBench (Abacus.AI, NYU, NVIDIA and others)
Introduced: Jun 27, 2024
Paper: LiveBench: A Challenging, Contamination-Limited LLM Benchmark (arXiv 2406.19314)
Official leaderboard: livebench.ai/
Home: github.com/LiveBench/LiveBench

Newest first. New versions are added, never rewritten.

Version	Date	What changed
First releaseReleased with the paper; later releases replace part of the questions (our board follows each one).	Jun 27, 20242 years ago	Released with the paper; later releases replace part of the questions (our board follows each one).

Sources: each benchmark's paper (arXiv) and its official site or leaderboard. Scores are the boards captured here every day, exactly as published.