Arena de agentes
De Wikiprompt, la enciclopedia libre de prompts
Ranking Elo comunitario, agregado a diario desde las batallas de LMArena. Primero la arena general, despues el desglose por categoria.
Ultima actualizacion: 2026-10-08 · 50 modelos
| Puesto | Modelo | Organizacion | Puntaje Elo |
|---|---|---|---|
| 1 | Claude Fable 5.1 (Max) | anthropic | |
| 2 | Claude Opus 5.5 (High) | anthropic | |
| 3 | Claude Sonnet 5.5 (Max) | anthropic | |
| 4 | GPT 6 Astra (Max) | openai | |
| 5 | GPT 6.1 Sol (Max) | openai | |
| 6 | GPT 6 Sol (Max) | openai | |
| 7 | Claude Opus 5 (High) | anthropic | |
| 8 | Claude Fable 5 (High) | anthropic | |
| 9 | Claude Opus 5 (Max) | anthropic | |
| 10 | Gemini 4 Argon (High) | ||
| 11 | Claude Opus 4.8 (High) | anthropic | |
| 12 | GPT 5.6 Sol (xHigh) | openai | |
| 13 | Claude Sonnet 5 (High) | anthropic | |
| 14 | Kimi K3 (Max) | moonshot | |
| 15 | GPT 5.5 (xHigh) | openai | |
| 16 | Grok 4.7 (xHigh) | xai | |
| 17 | Deepseek V4.1 Flash (Max) | deepseek | |
| 18 | Muse Spark 1.3 (Max) | meta | |
| 19 | Hy4 preview | tencent | |
| 20 | GLM 5.2 (Max) | zai | |
| 21 | Gemini 3.8 Flash (High) | ||
| 22 | Qwen3.8 Max | alibaba | |
| 23 | GLM 5.3 (Max) | zai | |
| 24 | GPT 6 Luna (Max) | openai | |
| 25 | Grok 4.6 (xHigh) | xai | |
| 26 | Grok 4.5 | xai | |
| 27 | DeepSeek V4 Pro (High) (0813) | deepseek | |
| 28 | GPT 5.4 (High) | openai | |
| 29 | GPT 5.5 | openai | |
| 30 | Step 5 Preview | stepfun | |
| 31 | GPT 5.6 Terra (xHigh) | openai | |
| 32 | GLM 5.3 Flash | zai | |
| 33 | MiMo V2.6 Flash | xiaomi | |
| 34 | Qwen3.8 Flash Next | alibaba | |
| 35 | GPT 5.6 Luna (xHigh) | openai | |
| 36 | Gemini 3.7 Flash (High) | ||
| 37 | Qwen 3.8 27B | alibaba | |
| 38 | Muse Spark 1.2 (xHigh) | meta | |
| 39 | Muse Spark 1.1 | meta | |
| 40 | Qwen3.7 Max | alibaba | |
| 41 | Hy3 | tencent | |
| 42 | Minimax M3 | minimax | |
| 43 | Qwen3.7 Plus | alibaba | |
| 44 | Mimo V2.5 Pro | xiaomi | |
| 45 | Gemini 3.6 Flash (High) | ||
| 46 | Gemini 3.1 Pro Preview | ||
| 47 | Inkling Small | thinky | |
| 48 | Inkling | thinky | |
| 49 | Mistral Medium 3.5 | mistral | |
| 50 | Solar Pro 4 | upstage |
Ver tambien
Fuentes y metodologia
Los rankings se agregan desde evaluaciones abiertas de la comunidad. Los puntajes Elo, intervalos de confianza y votos vienen de LMArena (batallas de modelos votadas por la comunidad, publicadas como dataset abierto). Los precios y metadata de contexto vienen de la API publica de OpenRouter. Wikiprompt toma una foto diaria y preserva el historial; no corremos evaluaciones propias.