La misma comparación a ciegas, pero en inglés. Se lee mejor junto a la selección de español: la distancia entre los dos puestos de un mismo modelo muestra cuánto pierde fuera de la lengua en la que más lo entrenaron.
| # | Modelo | Desarrollador | Arena Score, inglés | Acceso$ / 1M | Salida$ / 1M | Contextotokens |
|---|---|---|---|---|---|---|
| 1 | Claude Opus 4.6 | Anthropic | 1.505,8 1.501–1.510 | $5 | $25 | 1.000K |
| 2 | Claude Opus 5 ≈ | Anthropic | 1.500,16 1.482–1.519 | $5 | $25 | 1.000K |
| 3 | Claude Fable 5 ≈ | Anthropic | 1.498,82 1.491–1.507 | $10 | $50 | 1.000K |
| 4 | Claude Opus 4.7 ≈ | Anthropic | 1.487,52 1.482–1.493 | $5 | $25 | 1.000K |
| 5 | Gemini 3.6 Flash ≈ | Google DeepMind | 1.483,77 1.472–1.495 | $1,5 | $7,5 | 1.049K |
| 6 | Muse Spark 1.1 ≈ | Meta AI | 1.480,18 1.471–1.490 | $1,25 | $4,25 | 1.049K |
| 7 | Gemini 3 Pro ≈ | Google DeepMind | 1.479,64 1.474–1.485 | $1,6 | $9,6 | 1.049K |
| 8 | Gemini 3.5 Flash ≈ | Google DeepMind | 1.479,43 1.470–1.488 | $1,5 | $9 | 1.049K |
| 9 | Gemini 3.1 Pro Preview ≈ | Google DeepMind | 1.478,03 1.474–1.483 | $2 | $12 | 1.049K |
| 10 | GLM 5.1 ≈ | Zhipu AI / Z.ai | 1.477,83 1.472–1.484 | $0,06 | $0,22 | 205K |
| 11 | MiMo V2.5 Pro ≈ | Xiaomi | 1.477,62 1.472–1.483 | $0,44 | $0,87 | 1.050K |
| 12 | Kimi K3 ≈ | Moonshot AI (Kimi) | 1.477,38 1.461–1.493 | $2 | $8 | 1.049K |
| 13 | ERNIE 5.1 ≈ | Baidu (Ernie) | 1.476,98 1.471–1.483 | $0,75 | $3 | 119K |
| 14 | Claude Sonnet 4.6 ≈ | Anthropic | 1.473,19 1.468–1.478 | $3 | $15 | 1.000K |
| 15 | GLM 5.2 ≈ | Zhipu AI / Z.ai | 1.468,69 1.461–1.476 | $0,42 | $1,32 | 1.049K |
| 16 | Gemini 3 Flash ≈ | Google DeepMind | 1.465,02 1.459–1.471 | $0,4 | $2,4 | 1.049K |
| 17 | GPT-5.6 Sol ≈ | OpenAI | 1.460,56 1.450–1.471 | $5 | $30 | 1.050K |
| 18 | Claude 4.5 Opus ≈ | Anthropic | 1.460,01 1.456–1.464 | $5 | $25 | 200K |
| 19 | Kimi K2.6 ≈ | Moonshot AI (Kimi) | 1.458,16 1.452–1.464 | $0,22 | $1,14 | 262K |
| 20 | DeepSeek V4 Pro ≈ | DeepSeek | 1.457,95 1.452–1.463 | $0,44 | $0,87 | 1.049K |
| 21 | Claude Opus 4.8 ≈ | Anthropic | 1.457,59 1.451–1.464 | $5 | $25 | 1.000K |
| 22 | Qwen3.7 Plus ≈ | Alibaba (Qwen) | 1.457,42 1.450–1.464 | $0,5 | $3 | 1.000K |
| 23 | Gemini 2.5 Pro ≈ | Google DeepMind | 1.457,33 1.454–1.461 | $1,25 | $10 | 1.049K |
| 24 | Grok 4.20 (Reasoning) ≈ | xAI | 1.457,27 1.452–1.462 | $2 | $6 | 2.000K |
| 25 | GLM 5 ≈ | Zhipu AI / Z.ai | 1.456,77 1.451–1.463 | $0,48 | $1,9 | 205K |
| 26 | Grok 4.5 ≈ | xAI | 1.455,92 1.446–1.465 | $2 | $6 | 500K |
| 27 | GLM 4.7 ≈ | Zhipu AI / Z.ai | 1.454,08 1.446–1.462 | $0,15 | $0,8 | 205K |
| 28 | MiMo V2 Pro ≈ | Xiaomi | 1.454,04 1.448–1.460 | $0,44 | $0,87 | 1.049K |
| 29 | GLM 5V Turbo ≈ | Zhipu AI / Z.ai | 1.454,04 1.445–1.464 | $0,7 | $3,1 | 203K |
| 30 | Kimi K2.5 Thinking TEE ≈ | Moonshot AI (Kimi) | 1.453,91 1.449–1.458 | $0,3 | $1,9 | 256K |
| 31 | Grok 4.20 Multi Agent Beta 0309 ≈ | xAI | 1.452,15 1.447–1.457 | $2 | $6 | 2.000K |
| 32 | Hy3 ≈ | Tencent (Hunyuan) | 1.451,61 1.435–1.468 | $0,13 | $0,53 | 262K |
| 33 | Claude Sonnet 4.5 ≈ | Anthropic | 1.451,19 1.447–1.455 | $3 | $15 | 1.000K |
| 34 | Seed 2.0 Pro ≈ | ByteDance (Doubao) | 1.449,94 1.445–1.455 | $0,5 | $3 | 131K |
| 35 | Qwen3.6 Max Preview ≈ | Alibaba (Qwen) | 1.449,79 1.438–1.462 | $1,3 | $7,8 | 262K |
| 36 | MiniMax M3 ≈ | MiniMax | 1.449,67 1.443–1.456 | $0,3 | $1,2 | 1.049K |
| 37 | GLM 4.6 ≈ | Zhipu AI / Z.ai | 1.448,2 1.443–1.453 | $0,29 | $1,14 | 205K |
| 38 | Claude Sonnet 5 ≈ | Anthropic | 1.448,18 1.440–1.456 | $2 | $10 | 1.000K |
| 39 | Gemma 4 31B ≈ | Google DeepMind | 1.448,17 1.437–1.459 | $0,14 | $0,4 | 262K |
| 40 | MiMo V2.5 ≈ | Xiaomi | 1.447,46 1.442–1.453 | $0,14 | $0,28 | 1.050K |
| 41 | Grok 4.1 ≈ | xAI | 1.446,43 1.442–1.451 | $2 | $10 | 200K |
| 42 | Qwen3.6 Plus ≈ | Alibaba (Qwen) | 1.445,67 1.440–1.451 | $0,5 | $3 | 1.000K |
| 43 | GPT-5.6 Luna ≈ | OpenAI | 1.445,19 1.435–1.455 | $0,2 | $1,2 | 1.050K |
| 44 | GPT-5.2 Chat ≈ | OpenAI | 1.444,41 1.439–1.450 | $1,75 | $14 | 128K |
| 45 | Qwen3.5 397B-A17B ≈ | Alibaba (Qwen) | 1.444,26 1.440–1.449 | $0,6 | $3,6 | 262K |
| 46 | GPT-5.6 Terra ≈ | OpenAI | 1.443,56 1.433–1.454 | $2 | $12 | 1.050K |
| 47 | Gemma 4 26B-A4B ≈ | Google DeepMind | 1.442,26 1.431–1.454 | $0,05 | $0,29 | 262K |
| 48 | DeepSeek V4 Flash ≈ | DeepSeek | 1.442,23 1.437–1.448 | $0,14 | $0,28 | 1.049K |
| 49 | Longcat Flash Chat ≈ | Meituan | 1.440,51 1.432–1.449 | — | — | 131K |
| 50 | Qwen3 Max Preview ≈ | Alibaba (Qwen) | 1.440,21 1.434–1.446 | $1,2 | $6 | 256K |
| 51 | Mistral Large 3 ≈ | Mistral AI | 1.439,72 1.435–1.444 | $0,5 | $1,5 | 256K |
| 52 | DeepSeek V3.2 Exp ≈ | DeepSeek | 1.439,48 1.431–1.448 | $0,22 | $0,33 | 164K |
| 53 | Inkling ≈ | Thinking Machines Lab | 1.439,1 1.428–1.450 | $1,87 | $4,68 | 1.049K |
| 54 | MiMo V2 Omni ≈ | Xiaomi | 1.436,91 1.430–1.444 | $0,14 | $0,28 | 262K |
| 55 | Mistral Medium 3.1 ≈ | Mistral AI | 1.436,8 1.433–1.440 | $0,4 | $2 | 262K |
| 56 | DeepSeek V3.2 ≈ | DeepSeek | 1.435,55 1.431–1.440 | $0,28 | $0,4 | 164K |
| 57 | DeepSeek R1 0528 ≈ | DeepSeek | 1.434,45 1.427–1.442 | $0,25 | $0,25 | 164K |
| 58 | Mistral Medium 3.5 ≈ | Mistral AI | 1.433,77 1.425–1.442 | $1,5 | $7,5 | 262K |
| 59 | Gemini 3.5 Flash Lite ≈ | Google DeepMind | 1.433,44 1.421–1.446 | $0,3 | $2,5 | 1.049K |
| 60 | GLM 4.5 ≈ | Zhipu AI / Z.ai | 1.433,41 1.427–1.440 | $0,2 | $0,8 | 131K |
| 61 | ChatGPT 4o Latest ≈ | OpenAI | 1.432 1.428–1.436 | $5 | $15 | 128K |
| 62 | MiMo V2 Flash ≈ | Xiaomi | 1.431,13 1.426–1.436 | $0,14 | $0,28 | 262K |
| 63 | DeepSeek V3.1 ≈ | DeepSeek | 1.430,47 1.423–1.438 | $0,2 | $0,7 | 164K |
| 64 | Qwen3 VL 235B A22B Instruct ≈ | Alibaba (Qwen) | 1.430,42 1.422–1.439 | $0,4 | $1,6 | 262K |
| 65 | Claude 4.1 Opus ≈ | Anthropic | 1.428,72 1.425–1.433 | $15 | $75 | 200K |
| 66 | Qwen3.5 122B-A10B ≈ | Alibaba (Qwen) | 1.428,63 1.423–1.434 | $0,4 | $3,2 | 262K |
| 67 | Kimi K2 Thinking Turbo ≈ | Moonshot AI (Kimi) | 1.428,12 1.424–1.432 | $1,15 | $8 | 262K |
| 68 | MiniMax M2.7 ≈ | MiniMax | 1.427,74 1.422–1.433 | $0,3 | $1,2 | 205K |
| 69 | Qwen3 235B A22B Thinking-2507 ≈ | Alibaba (Qwen) | 1.427,11 1.418–1.436 | $0,1 | $0,1 | 262K |
| 70 | Qwen3 Next 80B A3B Instruct ≈ | Alibaba (Qwen) | 1.425,02 1.419–1.431 | $0,15 | $1,2 | 262K |
| 71 | Qwen3 235B A22B Instruct 2507 ≈ | Alibaba (Qwen) | 1.424,86 1.421–1.428 | $0,1 | $0,1 | 262K |
| 72 | DeepSeek V3.1 Terminus ≈ | DeepSeek | 1.422,96 1.409–1.437 | $0,21 | $0,79 | 164K |
| 73 | Qwen3.5 27B ≈ | Alibaba (Qwen) | 1.422,64 1.417–1.428 | $0,3 | $2,4 | 262K |
| 74 | Qwen3 Max 2025-09-23 ≈ | Alibaba (Qwen) | 1.420,37 1.412–1.429 | $0,86 | $3,43 | 258K |
| 75 | GPT 4.5 Preview ≈ | OpenAI | 1.419,49 1.412–1.427 | $75 | $150 | 128K |
| 76 | Grok 4 ≈ | xAI | 1.418,4 1.413–1.423 | $3 | $15 | 256K |
| 77 | Step 3.5 Flash ≈ | StepFun | 1.418,24 1.413–1.423 | $0,1 | $0,3 | 262K |
| 78 | GPT-5.5 Instant ≈ | OpenAI | 1.418,15 1.412–1.425 | $5 | $30 | 400K |
| 79 | Gemini 2.5 Flash ≈ | Google DeepMind | 1.418,1 1.415–1.421 | $0,3 | $2,5 | 1.049K |
| 80 | Grok 4.1 Fast Reasoning ≈ | xAI | 1.417,98 1.414–1.422 | $0,2 | $0,5 | 2.000K |
| 81 | Qwen3 VL 235B A22B Thinking ≈ | Alibaba (Qwen) | 1.417,86 1.408–1.427 | $0,4 | $4 | 262K |
| 82 | o3 2025-04-16 ≈ | OpenAI | 1.414,47 1.410–1.419 | $2 | $8 | 200K |
| 83 | Claude Haiku 4.5 ≈ | Anthropic | 1.413,97 1.410–1.418 | $1 | $5 | 200K |
| 84 | Gemini 3.1 Flash Lite Preview ≈ | Google DeepMind | 1.412,54 1.408–1.417 | $0,25 | $1,5 | 1.049K |
| 85 | 2025) ≈ | Google DeepMind | 1.411,08 1.406–1.416 | $0,3 | $2,5 | 1.049K |
| 86 | MiMo V2 Flash (Thinking) ≈ | Xiaomi | 1.410,51 1.402–1.419 | $0,1 | $0,31 | 256K |
| 87 | Qwen3.5 35B A3B ≈ | Alibaba (Qwen) | 1.407,94 1.402–1.414 | $0,25 | $2 | 262K |
| 88 | GPT 5 Chat ≈ | OpenAI | 1.407,16 1.402–1.413 | $1,25 | $10 | 128K |
| 89 | Grok 4 Fast Reasoning ≈ | xAI | 1.407,02 1.400–1.414 | $0,2 | $0,5 | 2.000K |
| 90 | Grok 4.3 ≈ | xAI | 1.405,67 1.400–1.411 | $1,25 | $2,5 | 1.000K |
| 91 | Qwen3.5 Flash ≈ | Alibaba (Qwen) | 1.405,65 1.401–1.411 | $0,09 | $0,36 | 1.000K |
| 92 | Nemotron 3 Super ≈ | NVIDIA | 1.400,01 1.390–1.410 | $0,2 | $0,8 | 1.000K |
| 93 | Hunyuan T1 ≈ | Tencent (Hunyuan) | 1.395,04 1.382–1.408 | — | — | 131K |
| 94 | GLM 4.5 Air ≈ | Zhipu AI / Z.ai | 1.394,99 1.390–1.400 | $0,11 | $0,29 | 131K |
| 95 | Qwen3 Next 80B A3B Thinking ≈ | Alibaba (Qwen) | 1.393,03 1.385–1.401 | $0,15 | $1,2 | 262K |
| 96 | GLM 4.6V ≈ | Zhipu AI / Z.ai | 1.392,39 1.376–1.409 | $0,14 | $0,42 | 131K |
| 97 | Qwen3 30B A3B Instruct 2507 ≈ | Alibaba (Qwen) | 1.391,23 1.385–1.397 | $0,05 | $0,19 | 262K |
| 98 | GPT 4.1 2025-04-14 ≈ | OpenAI | 1.390,3 1.386–1.395 | $2 | $8 | 1.048K |
| 99 | GPT-5.3 Chat (latest) ≈ | OpenAI | 1.389,83 1.384–1.396 | $1,75 | $14 | 128K |
| 100 | DeepSeek Chat 0324 ≈ | DeepSeek | 1.386,46 1.382–1.391 | $0,2 | $0,6 | 164K |
La tabla se desplaza en horizontal: no caben todas las columnas.
El precio es el más bajo entre los proveedores del modelo, tarifa base, sin tarifas por lotes ni reducidas. La entrada y la salida se muestran por separado a propósito: en la mayoría de los modelos la salida cuesta varias veces más que la entrada, y la factura final depende de qué predomine en la tarea.
El signo ≈ marca los modelos cuyo intervalo de confianza se solapa con el de la fila de arriba. Aquí hay 99 modelos — su orden entre sí no está determinado por los datos disponibles.
El intervalo de confianza de la mitad de los modelos supera 11,6 puntos — es decir, unas 10 filas adyacentes de la tabla. Dentro de ese grupo el orden no lo marcan los datos, sino quién votó esta vez; los datos sostienen la diferencia entre el principio y el final de la lista, pero no la vecindad de los puestos. Votos por modelo aquí — mediana 13.842: cuantos menos hay, más ancho es el intervalo.
Arena (LMArena)
— licencia CC-BY-4.0 · fuente original
По данным Arena (LMArena)