SWE-rebench Leaderboard

1AnthropicFable 5 [high]Model64.5%± 1.41%78.4%$4.402,518,30894.9% cached2GrokGrok 4.5 [high]Model63.8%± 0.60%77.5%$1.472,429,42492.6% cached3AnthropicOpus 5 [high]Model63.4%± 1.35%74.8%$3.474,322,14395.7% cached4Z.aiGLM-5.2 [high]Model62.9%± 1.19%81.1%$1.405,524,89292.1% cached5OpenAIGPT-5.6 Sol [medium]Model62.3%± 1.83%79.3%$0.85605,34084.7% cached6JunieJunieAgent61.8%± 0.54%73.9%$0.811,684,50185.9% cached7AnthropicClaude CodeAgent60.4%± 1.03%75.7%$3.393,341,58193.4% cached8OpenAICodexAgent58.0%± 1.29%73.0%$1.592,070,97694.6% cached9AnthropicSonnet 5 [high]Model56.8%± 0.94%74.8%$1.434,645,61796.4% cached10CursorCursorAgent51.7%± 0.84%65.8%$0.411,827,00294.3% cached11MinimaxMiniMax M3Model47.2%± 1.13%69.4%$0.9513,869,45997.0% cached12XiaomiMiMoMiMo V2.5 ProModel46.5%± 0.54%65.8%$0.104,687,98795.7% cached13OpenAIGPT-5.6 Luna [medium]Model43.6%± 1.47%59.5%$0.11395,52285.2% cached14DeepSeekDeepSeek-V4 Pro [high]Model40.2%± 1.29%64.0%$0.153,955,41491.2% cached15QwenQwen3.6-27BModel31.2%± 1.68%57.7%$0.623,078,97076.8% cached16QwenQwen3.6-35B-A3BModel24.7%± 0.79%43.2%$0.273,726,06187.5% cached17QwenQwen3.5-35B-A3BModel17.1%± 1.07%36.9%$0.995,939,22772.3% cached18AnthropicClaude Opus 4.1ModelN/AN/AN/AN/A19AnthropicClaude Opus 4.5ModelN/AN/AN/AN/A20AnthropicClaude Opus 4.6-highModelN/AN/AN/AN/A21AnthropicClaude Opus 4.7-highModelN/AN/AN/AN/A22AnthropicClaude Opus 4.8-xhighModelN/AN/AN/AN/A23AnthropicClaude Sonnet 3.5ModelN/AN/AN/AN/A24AnthropicClaude Sonnet 4ModelN/AN/AN/AN/A25AnthropicClaude Sonnet 4.5ModelN/AN/AN/AN/A26AnthropicClaude Sonnet 4.6ModelN/AN/AN/AN/A27DeepSeekDeepSeek-R1-0528ModelN/AN/AN/AN/A28DeepSeekDeepSeek-V3ModelN/AN/AN/AN/A29DeepSeekDeepSeek-V3-0324ModelN/AN/AN/AN/A30DeepSeekDeepSeek-V3-0324ModelN/AN/AN/AN/A31DeepSeekDeepSeek-V3.1ModelN/AN/AN/AN/A32DeepSeekDeepSeek-V3.2ModelN/AN/AN/AN/A33DeepSeekDeepSeek-V4 Flash [high]ModelN/AN/AN/AN/A34MistralDevstral-2-123B-Instruct-2512ModelN/AN/AN/AN/A35MistralDevstral-Small-2-24B-Instruct-2512ModelN/AN/AN/AN/A36MistralDevstral-Small-2505ModelN/AN/AN/AN/A37GeminiGemini 3 Flash PreviewModelN/AN/AN/AN/A38GeminiGemini 3 Pro PreviewModelN/AN/AN/AN/A39GeminiGemini 3.1 Pro PreviewModelN/AN/AN/AN/A40GeminiGemini 3.5 FlashModelN/AN/AN/AN/A41Geminigemini-2.0-flashModelN/AN/AN/AN/A42Geminigemini-2.0-flashModelN/AN/AN/AN/A43Geminigemini-2.5-flashModelN/AN/AN/AN/A44Geminigemini-2.5-flash-preview-05-20 no-thinkingModelN/AN/AN/AN/A45Geminigemini-2.5-flash-preview-05-20 no-thinkingModelN/AN/AN/AN/A46Geminigemini-2.5-proModelN/AN/AN/AN/A47GeminiGemma 4 31BModelN/AN/AN/AN/A48Geminigemma-3-27b-itModelN/AN/AN/AN/A49Z.aiGLM-4.5ModelN/AN/AN/AN/A50Z.aiGLM-4.5 AirModelN/AN/AN/AN/A51Z.aiGLM-4.6ModelN/AN/AN/AN/A52Z.aiGLM-4.7ModelN/AN/AN/AN/A53Z.aiGLM-4.7 FlashModelN/AN/AN/AN/A54Z.aiGLM-5ModelN/AN/AN/AN/A55Z.aiGLM-5.1ModelN/AN/AN/AN/A56Z.aiGLM-5.1ModelN/AN/AN/AN/A57OpenAIgpt-4.1-2025-04-14ModelN/AN/AN/AN/A58OpenAIgpt-4.1-2025-04-14ModelN/AN/AN/AN/A59OpenAIgpt-4.1-mini-2025-04-14ModelN/AN/AN/AN/A60OpenAIgpt-4.1-mini-2025-04-14ModelN/AN/AN/AN/A61OpenAIgpt-4.1-nano-2025-04-14ModelN/AN/AN/AN/A62OpenAIgpt-5-2025-08-07-highModelN/AN/AN/AN/A63OpenAIgpt-5-2025-08-07-mediumModelN/AN/AN/AN/A64OpenAIgpt-5-2025-08-07-minimalModelN/AN/AN/AN/A65OpenAIgpt-5-codexModelN/AN/AN/AN/A66OpenAIgpt-5-mini-2025-08-07-highModelN/AN/AN/AN/A67OpenAIgpt-5-mini-2025-08-07-mediumModelN/AN/AN/AN/A68OpenAIgpt-5.1-codexModelN/AN/AN/AN/A69OpenAIgpt-5.1-codex-maxModelN/AN/AN/AN/A70OpenAIgpt-5.2-2025-12-11-mediumModelN/AN/AN/AN/A71OpenAIgpt-5.2-2025-12-11-xhighModelN/AN/AN/AN/A72OpenAIgpt-5.2-codexModelN/AN/AN/AN/A73OpenAIgpt-5.3-codexModelN/AN/AN/AN/A74OpenAIgpt-5.3-codex-xhighModelN/AN/AN/AN/A75OpenAIgpt-5.4-2026-03-05-mediumModelN/AN/AN/AN/A76OpenAIgpt-5.5-2026-04-23-mediumModelN/AN/AN/AN/A77OpenAIgpt-5.5-2026-04-23-xhighModelN/AN/AN/AN/A78OpenAIgpt-oss-120bModelN/AN/AN/AN/A79OpenAIgpt-oss-120b-highModelN/AN/AN/AN/A80OpenAIgpt-oss-20bModelN/AN/AN/AN/A81GrokGrok 4ModelN/AN/AN/AN/A82GrokGrok Code Fast 1ModelN/AN/AN/AN/A83OpenRouterhorizon-alphaModelN/AN/AN/AN/A84OpenRouterhorizon-betaModelN/AN/AN/AN/A85KimiKimi K2ModelN/AN/AN/AN/A86KimiKimi K2 Instruct 0905ModelN/AN/AN/AN/A87KimiKimi K2 ThinkingModelN/AN/AN/AN/A88KimiKimi K2.5ModelN/AN/AN/AN/A89KimiKimi K2.6ModelN/AN/AN/AN/A90MetaLlama-3.3-70B-InstructModelN/AN/AN/AN/A91MetaLlama-4-Maverick-17B-128E-InstructModelN/AN/AN/AN/A92MetaLlama-4-Scout-17B-16E-InstructModelN/AN/AN/AN/A93MinimaxMiniMax M2ModelN/AN/AN/AN/A94MinimaxMiniMax M2.1ModelN/AN/AN/AN/A95MinimaxMiniMax M2.5ModelN/AN/AN/AN/A96MinimaxMiniMax M2.7ModelN/AN/AN/AN/A97OpenAIo3-2025-04-16ModelN/AN/AN/AN/A98OpenAIo4-mini-2025-04-16ModelN/AN/AN/AN/A99QwenQwen2.5-72B-InstructModelN/AN/AN/AN/A100QwenQwen2.5-Coder-32B-InstructModelN/AN/AN/AN/A101QwenQwen3-235B-A22BModelN/AN/AN/AN/A102QwenQwen3-235B-A22B no-thinkingModelN/AN/AN/AN/A103QwenQwen3-235B-A22B thinkingModelN/AN/AN/AN/A104QwenQwen3-235B-A22B-Instruct-2507ModelN/AN/AN/AN/A105QwenQwen3-235B-A22B-Thinking-2507ModelN/AN/AN/AN/A106QwenQwen3-30B-A3B-Instruct-2507ModelN/AN/AN/AN/A107QwenQwen3-30B-A3B-Thinking-2507ModelN/AN/AN/AN/A108QwenQwen3-32BModelN/AN/AN/AN/A109QwenQwen3-32B no-thinkingModelN/AN/AN/AN/A110QwenQwen3-32B thinkingModelN/AN/AN/AN/A111QwenQwen3-Coder-30B-A3B-InstructModelN/AN/AN/AN/A112QwenQwen3-Coder-480B-A35B-InstructModelN/AN/AN/AN/A113QwenQwen3-Coder-NextModelN/AN/AN/AN/A114QwenQwen3-Next-80B-A3B-InstructModelN/AN/AN/AN/A115QwenQwen3.5-27BModelN/AN/AN/AN/A116QwenQwen3.5-397B-A17BModelN/AN/AN/AN/A117StepfunStep-3.5-FlashModelN/AN/AN/AN/A