Charts v2

Score distribution v2

Interactive v2 model comparison using the same JSON result files that power the leaderboard.

Average total scoreScale: 0 to 100

Claude Sonnet 4.6Anthropic / hosted89.3 Claude Haiku 4.5, Claude web free planAnthropic / hosted83.7 Qwen 3.6 Thinking, Ollama low-thinkingQwen / local83 Gemini 3.1 Pro, Gemini web standardGoogle / hosted82 Gemini 3.5 Flash, Gemini web free/defaultGoogle / hosted64.7 gemma4:e2bOllama/local model source unknown / local59