OpenCodePapers

code-generation-on-swe-marathon

Code IntelligenceCode Generation
Dataset Link
Results over time
Click legend items to toggle metrics. Hover points for model names.
Leaderboard
PaperCodeResolution RateModelNameReleaseDate
SWE-Marathon Leaderboard53.12Muse Wonder Full (max) (Muse Code)2026-10-12
SWE-Marathon Leaderboard52.5Muse Wonder (max) (Muse Code)2026-10-12
SWE-Marathon Leaderboard✓ Link51.25Claude Sonnet 5.5 (max) (Claude Code)2026-09-28
SWE-Marathon Leaderboard✓ Link50.0Claude Opus 5 (max) (Claude Code)2026-07-24
SWE-Marathon Leaderboard50.0Claude Haiku 5.5 (max) (Claude Code)2026-10-07
SWE-Marathon Leaderboard48.75Claude Opus 4.8 (Claude Code)2026-05-28
SWE-Marathon Leaderboard✓ Link48.13Kimi K3 (Claude Code)2026-07-27
SWE-Marathon Leaderboard✓ Link46.25Claude Opus 5.5 (max) (Claude Code)2026-09-22
SWE-Marathon Leaderboard45.62Claude Fable 5.1 (max) (Claude Code)2026-09-01
SWE-Marathon Leaderboard45.62Grok 4.7 (high) (Grok Build)2026-09-21
SWE-Marathon Leaderboard43.12GPT-6 Sol (max) (Codex)2026-09-22
SWE-Marathon Leaderboard✓ Link42.5GPT-5.6 Sol (max) (Codex)2026-07-09
SWE-Marathon Leaderboard✓ Link42.5GLM-5.3 (max) (Claude Code)2026-08-14
SWE-Marathon Leaderboard✓ Link42.5GPT-6 Astra (max) (Codex)2026-09-03
SWE-Marathon Leaderboard42.5GPT-6.1 Sol (max) (Codex)2026-09-29
SWE-Marathon Leaderboard✓ Link38.75GLM-5.3-Flash (max) (Claude Code)2026-08-26
SWE-Marathon Leaderboard✓ Link32.5GPT-5.6 Terra (max) (Codex)2026-07-09
SWE-Marathon Leaderboard31.88Grok 4.6 (high) (Grok Build)2026-08-12
SWE-Marathon Leaderboard30.62Muse Spark 1.3 (xhigh) (Muse Code)2026-09-02
SWE-Marathon Leaderboard30.0Claude Sonnet 5 (max) (Claude Code)2026-06-30
SWE-Marathon Leaderboard29.38Grok 4.5 (Grok Build)2026-07-08
SWE-Marathon Leaderboard28.12GPT-6 Luna (max) (Codex)2026-09-22
SWE-Marathon Leaderboard✓ Link24.38GPT-5.6 Luna (max) (Codex)2026-07-09
SWE-Marathon Leaderboard✓ Link22.5Gemini 3.7 Flash (OpenCode)2026-08-13
SWE-Marathon Leaderboard✓ Link19.38GLM-5.2 (Claude Code)2026-06-16
SWE-Marathon Leaderboard✓ Link10.63DeepSeek V4 Pro (MiniSWE-Agent)2026-04-26