Models
Ranked by Real-SWE resolution rate
Claude Opus 5.5Rank 1Resolution rate: 56.80%
GPT 6 AstraRank 2Resolution rate: 56.00%
Gemini 4 CarbonRank 3Resolution rate: 55.20%Claude Sonnet 5.5Rank 4Resolution rate: 51.20%
GPT 6.1 SolRank 5Resolution rate: 47.20%
Claude Fable 5.1Rank 6Resolution rate: 45.60%
Muse Wonder 1.0Rank 7Resolution rate: 42.40%
Gemini 4 ArgonTied for 8Resolution rate: 39.20%Kimi K3.1 ProTied for 8Resolution rate: 39.20%
Particle-1Tied for 10Resolution rate: 38.40%
DeepSeek V4.1 FlashTied for 10Resolution rate: 38.40%
Hy4 previewRank 12Resolution rate: 33.60%
Inkling 2 SmallRank 13Resolution rate: 32.00%
Grok 4.7Rank 14Resolution rate: 28.00%
GLM-5.3Rank 15Resolution rate: 26.40%
Qwen3.8-Max primeRank 16Resolution rate: 24.00%
GPT 6 LunaRank 17Resolution rate: 10.40%
Inkling 1Rank 18Resolution rate: 8.00%
Seed 2.1 TurboRank 19Resolution rate: 5.60%
Nemotron 3 UltraRank 20Resolution rate: 4.80%
About Specific
Specific is an independent evaluator of AI model capabilities, with a focus on coding. Real-SWE measures how models perform on real engineering tasks in private, production codebases.
