Página automatizada. Redactada y traducida automáticamente por IA y verificada con las fuentes mediante software; sin editor humano. SI-IQ es un índice elaborado a partir de resultados públicos de benchmarks, no una prueba de CI. Cómo funciona

SWE-bench Verified (Epoch AI run): resultados de modelos de IA

SWE-bench Verified (Epoch AI run) es un benchmark de Programación ejecutado por Epoch AI. 32 modelos de nuestros datos tienen un resultado publicado.

Forma parte del índice SI-IQ (carga k = 0,47, nivel de azar 0 %).

Evaluador
Epoch AI
Página principal del benchmark
openai.com/index/introducing-swe-bench-verified

Resultados

Fuente
1 Claude Opus 4.7 Anthropic83,5 %127124–130 9 abr 2026Epoch AI Benchmarking Hub
2 GPT-5.5 OpenAI80,6 %138135–141 9 abr 2026Epoch AI Benchmarking Hub
3 Gemini 3.5 Flash Google DeepMind79,3 %126123–129 9 may 2026Epoch AI Benchmarking Hub
4 Claude Opus 4.6 Anthropic78,7 %123120–126 9 feb 2026Epoch AI Benchmarking Hub
5 GLM-5.2abierto Z.ai (Zhipu AI)78,7 %121118–124 9 jun 2026Epoch AI Benchmarking Hub
6 DeepSeek-V4-Proabierto DeepSeek77,6 %117113–120 9 abr 2026Epoch AI Benchmarking Hub
7 Qwen3.7-Max Alibaba77,3 %125122–128 9 may 2026Epoch AI Benchmarking Hub
8 GPT-5.4 OpenAI76,9 %131128–133 9 mar 2026Epoch AI Benchmarking Hub
9 Claude Opus 4.5 Anthropic76,7 %115111–118 10 nov 2025Epoch AI Benchmarking Hub
10 Kimi K2.6abierto Moonshot76,7 %121118–124 9 abr 2026Epoch AI Benchmarking Hub
11 Qwen 3.6 Max (Preview) Alibaba76,7 %118112–124 5 datos insuficientes abr 2026Epoch AI Benchmarking Hub
12 Gemini 3.1 Pro Google DeepMind75,6 %127124–130 9 feb 2026Epoch AI Benchmarking Hub
13 Gemini 3 Flash Google DeepMind75,4 %122119–125 9 dic 2025Epoch AI Benchmarking Hub
14 Claude Sonnet 4.6 Anthropic75,2 %113108–118 7 feb 2026Epoch AI Benchmarking Hub
15 GPT-5.3 Codex OpenAI74,8 %122108–136 2 datos insuficientes feb 2026Epoch AI Benchmarking Hub
16 GLM-5.1abierto Z.ai (Zhipu AI)74,2 %116112–120 7 abr 2026Epoch AI Benchmarking Hub
17 GPT-5.2 OpenAI73,8 %125122–128 10 dic 2025Epoch AI Benchmarking Hub
18 Kimi K2.5abierto Moonshot73,8 %119112–125 5 ene 2026Epoch AI Benchmarking Hub
19 GPT-5 OpenAI73,6 %121118–124 11 ago 2025Epoch AI Benchmarking Hub
20 Claude Opus 4.1 Anthropic73,3 %105102–109 6 datos insuficientes ago 2025Epoch AI Benchmarking Hub
21 Gemini 3 Pro Google DeepMind72,9 %122115–130 4 datos insuficientes nov 2025Epoch AI Benchmarking Hub
22 GLM-5abierto Z.ai (Zhipu AI)72,1 %112105–118 4 datos insuficientes feb 2026Epoch AI Benchmarking Hub
23 Claude Sonnet 4.5 Anthropic71,3 %110107–113 11 sept 2025Epoch AI Benchmarking Hub
24 Claude Opus 4 Anthropic70,7 %9893–102 5 may 2025Epoch AI Benchmarking Hub
25 GPT-5.1 OpenAI68,0 %120115–125 7 nov 2025Epoch AI Benchmarking Hub
26 GPT-5 mini OpenAI64,7 %115112–118 11 ago 2025Epoch AI Benchmarking Hub
27 o3 OpenAI62,3 %116113–119 10 abr 2025Epoch AI Benchmarking Hub
28 Claude 3.7 Sonnet Anthropic61,0 %9995–103 5 feb 2025Epoch AI Benchmarking Hub
29 Qwen 3.6 Plus Alibaba57,9 %113109–117 7 datos insuficientes mar 2026Epoch AI Benchmarking Hub
30 Gemini 2.5 Pro (Jun 2025) Google DeepMind57,6 %108104–112 7 jun 2025Epoch AI Benchmarking Hub
31 GPT-4.1 OpenAI48,5 %9187–94 10 abr 2025Epoch AI Benchmarking Hub
32 GPT-4o (Nov 2024) OpenAI31,0 %6964–73 5 nov 2024Epoch AI Benchmarking Hub

Fuentes de datos: Aider polyglot leaderboard data (Apache-2.0); Arena leaderboard dataset on Hugging Face (CC BY 4.0); Berkeley Function Calling Leaderboard, Gorilla project (Apache-2.0); Epoch AI, ‘Capabilities & benchmarking’ (CC BY 4.0); API list prices via LiteLLM's price table (MIT); each price links to the developer's pricing page; Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter..