Share your thoughts, 1 month free Claude Pro on usSee more

Coding Agent Capacity on SWE-Bench Pro

55.6Score

GPT-5.2 High

Updated 24d ago

Evaluation Results

Method	Links
GPT-5.2 High 2026.06		55.6
Claude-Opus-4.5 2026.06		55.4
GPT-5-mini High 2026.06		51.7
Gemini-3-pro High 2026.06		49.7
Claude-Sonnet-4.5 2026.06		48.4
Seed2.0 Pro 2026.06		46.9
Gemini-3-Flash High 2026.06		46.7
Seed2.0 Lite 2026.06		46