Preference
Prompt-to-Leaderboard
Prompt-to-Leaderboard: prompt-adaptive LLM leaderboards from Chatbot Arena human pairwise battles (win/loss/tie per prompt).
128,287items
53subjects
4%observed
CC-BY-4.0license
preferencedomain
textmodality
Response matrix
Fit to width. Hover for subject & item; click a cell for details.
lowhighUnobserved
Scale: {0, 0.5, 1} (loss / tie / win, human pairwise vote)
Subjects
- 1gemini-2.5-pro0.6711
- 2gemini-2.5-pro-preview-03-250.635
- 3grok-4-07090.6291
- 4o3-2025-04-160.6099
- 5chatgpt-4o-latest-202503260.6061
- 6deepseek-r1-05280.5988
- 7gemini-2.5-pro-preview-05-060.5968
- 8grok-3-preview-02-240.594
- 9llama-4-maverick-03-26-experimental0.5881
- 10gemini-2.5-flash0.5847
- 11qwen3-235b-a22b-no-thinking0.5559
- 12gemini-2.5-flash-preview-04-170.5464
- 13kimi-k2-0711-preview0.5445
- 14gemini-2.0-flash-thinking-exp-01-210.5437
- 15gpt-4.1-2025-04-140.5315
- 16deepseek-v3-03240.5302
- 17hunyuan-turbos-202504160.5281
- 18o4-mini-2025-04-160.5266
- 19gemini-2.5-flash-lite-preview-06-17-thinking0.5256
- 20mistral-medium-25050.5252
- 21qwen3-235b-a22b0.5248
- 22qwen-max-2025-01-250.5235
- 23qwen3-235b-a22b-instruct-25070.5201
- 24minimax-m10.5166
- 25gemma-3-27b-it0.5096
- 26claude-opus-4-20250514-thinking-16k0.505
- 27grok-3-mini-high0.5043
- 28grok-3-mini-beta0.5005
- 29qwen3-coder-480b-a35b-instruct0.5
- 30claude-opus-4-202505140.4986
- 31gpt-4o-2024-11-200.4948
- 32claude-sonnet-4-20250514-thinking-32k0.479
- 33gemini-2.0-flash-0010.4772
- 34mistral-small-25060.4749
- 35qwq-32b0.4715
- 36gpt-4.1-mini-2025-04-140.4689