rank,harness,model,score,formal,end_result,instruction_compliance,pitfalls,pleasantness,tool_calls
1,Explyt AI Agent,Opus 4.7,81.5,81.2,94.0,75.0,64.0,92.0,83.0
2,Claude Code,Opus 4.7 (xhigh effort),76.2,81.2,91.0,69.0,58.0,78.0,80.0
3,Explyt AI Agent,GPT-5.5,73.0,75.0,78.0,70.0,55.0,80.0,80.0
4,Explyt AI Agent,Sonnet 4.6,70.2,81.2,66.0,62.0,56.0,73.0,83.0
5,Claude Code,Sonnet 4.6,70.1,90.6,80.0,62.0,55.0,75.0,58.0
6,Explyt AI Agent,GLM-5.1 (self-hosted),67.3,72.5,71.8,57.4,53.2,75.8,73.0
7,Explyt AI Agent,DeepSeek V4 Flash,64.8,87.5,73.0,50.0,50.0,66.0,62.0
8,Explyt AI Agent,DeepSeek V4 Pro,64.1,87.5,66.0,53.0,48.0,69.0,61.0
9,Claude Code,Haiku 4.5,63.6,84.4,78.0,58.0,48.0,58.0,55.0
10,Explyt AI Agent,GLM-5.1,60.6,80.6,60.0,53.0,48.0,61.0,61.0
11,Explyt AI Agent,Haiku 4.5,58.2,75.0,72.0,55.0,47.0,47.0,53.0
12,Explyt AI Agent,Gemini 3.1 Pro Preview,54.5,71.9,62.0,52.0,41.0,50.0,50.0
13,Explyt AI Agent,Gemini 3 Flash,52.7,75.0,66.0,48.0,41.0,39.0,47.0
14,Explyt AI Agent,MiniMax 2.7,50.8,70.0,52.0,47.0,38.0,43.0,55.0
15,Explyt AI Agent,Qwen 3.6 Plus,49.1,65.6,58.0,45.0,39.0,42.0,45.0
16,Explyt AI Agent,Mimo V2.5 Pro,45.1,62.5,50.0,38.0,32.0,36.0,52.0
17,Explyt AI Agent,Kimi K2.6*,28.1,42.9,29.0,33.0,19.0,21.0,24.0