List your company

Instruction Following on MARS-Bench

87.9ScoreGPT-5.2 High
61.3868.26575.1582.035Jun 30, 2026

Evaluation results

MethodLinks
GPT-5.2 High2026.0687.9
Claude-Opus-4.52026.0687.7
Gemini-3-pro High2026.0685.6
Seed2.0 Pro2026.0685.6
Gemini-3-FlashTier=High2026.0684.6
Seed2.0 Lite2026.0680.5
Claude-Sonnet-4.52026.0672.9
GPT-5-miniTier=High2026.0666.1
Seed2.0 Mini2026.0662.4