benchmarks.site

Public leaderboard · General knowledge & reasoning

LiveBench

A set of regularly refreshed tests (agent tasks, reasoning, coding, research, design, etc.) designed so AI models cannot have seen the questions beforehand.

Leaderboard

Top 3 right now

  1. 1stclaude-fable-5.284.9
  2. 2ndgpt-6-astra84.4
  3. 3rdclaude-fable-5.184.0

What it measures

Regularly refreshed questions in math, coding, reasoning, language, data analysis and instruction following, scored automatically against objective ground-truth answers.

How to read the numbers

Weighted overall score across LiveBench 2.0 categories (0-100) — higher is better.

Results

25 rows · higher is better (Overall (computed))CSV
1claude-fable-5.284.982.988.879.2
2gpt-6-astra84.475.692.787.5
3claude-fable-5.184.074.291.789.7
4gemini-4-carbon83.572.790.691.0
5claude-opus-5.583.472.892.287.1
6gpt-6.1-sol81.465.692.682.1
7gpt-6.1-terra78.856.288.789.2
8gemini-4-argon77.758.595.185.7
9gemini-4-hydrogen77.061.888.678.5
10claude-sonnet-5.577.063.491.677.2
11qwen4-max73.347.592.269.8
12muse-wonder-1.070.746.290.674.9
13deepseek-v4.1-flash-101770.338.889.180.4
14claude-haiku-5.568.947.982.477.6
15kimi-k3.1-pro68.738.791.167.8
16muse-spark-1.368.647.989.784.4
17grok-4.767.945.382.773.1
18gpt-5.6-terra67.225.990.684.2
19mimo-v2.6-pro66.136.479.664.5
20kimi-k365.036.090.753.1
21minimax-m3.1-flash63.725.782.968.8
22grok-4.663.518.791.276.7
23qwen3.8-max60.926.688.260.9
24glm-5.3-flash60.720.777.670.9
25gpt-6-luna60.424.781.870.8

Note: Release 2026-11-06 of LiveBench 2.0. Site publishes only category scores in table.csv; Overall computed here as weighted mean using the site categories.json weights (Long Agentic 2, Coding 1.5, Design 1.5, others 1) and may differ slightly from the site display. Also has Browser, Data Analysis, Research, Design columns in source.

Source: LiveBench team (Abacus.AI and collaborators) · livebench.ai · results as of 2026-11-06 · retrieved 2026-11-17

Markdown · More General knowledge & reasoning