Last updated:
Explorar modelos por modalidad
Ranking global (todas las categorías), la tabla de modelos abajo y atajos por modalidad (LLM, imagen, video, multimodal).
Ranking de modelos (todos)
Ranking de modelos
Rendimiento entre tareas (multimodal/visión/lenguaje; datos de ejemplo).
Public ranking policy: rows are sorted by composite score (desc). Composite score is a weighted sum of normalized sub-metrics; ties are broken by higher recent activity.
| Puesto | Modelo | Proveedor / equipo | Tipo | Puntuación | Notas |
|---|---|---|---|---|---|
| 1 | xAI: Grok 4.1 Fast | xAI | Multimodal | 98.7 | 2.0M ctx · $0.35/1M avg |
| 2 | OpenAI: GPT-5.4 | OpenAI | Multimodal | 62 | 1.1M ctx · $8.75/1M avg |
| 3 | Google: Lyria 3 Pro Preview | Multimodal | 62 | 1.0M ctx · $0.00/1M avg | |
| 4 | Meta: Llama 4 Maverick | Meta | Multimodal | 72.4 | 1.0M ctx · $0.38/1M avg |
| 5 | MiniMax: MiniMax-01 | MiniMax | Multimodal | 69.7 | 1.0M ctx · $0.65/1M avg |
| 6 | Qwen: Qwen3.5-Flash | Qwen | Multimodal | 72.2 | 1.0M ctx · $0.16/1M avg |
| 7 | Amazon: Nova 2 Lite | Amazon | Multimodal | 65.9 | 1.0M ctx · $1.40/1M avg |
| 8 | Anthropic: Claude Sonnet 4.6 | Anthropic | Multimodal | 62 | 1.0M ctx · $9.00/1M avg |
| 9 | Mistral: Ministral 3 8B 2512 ≈5 GB VRAM (estimación baja de inferencia 4-bit) | Mistral | Multimodal | 62 | 262k ctx · $0.15/1M avg |
| 10 | ByteDance Seed: Seed 1.6 Flash | ByteDance Seed | Multimodal | 62 | 262k ctx · $0.19/1M avg |
| 11 | MoonshotAI: Kimi K2.5 | MoonshotAI | Multimodal | 62 | 262k ctx · $1.05/1M avg |
| 12 | Xiaomi: MiMo-V2-Omni | Xiaomi | Multimodal | 62 | 262k ctx · $1.20/1M avg |
| 13 | Z.ai: GLM 5V Turbo | Z.ai | Multimodal | 62 | 203k ctx · $2.60/1M avg |
| 14 | Free Models Router | Free | Multimodal | 62 | 200k ctx · $0.00/1M avg |
| 15 | Perplexity: Sonar Pro Search | Perplexity | Multimodal | 62 | 200k ctx · $9.00/1M avg |
| 16 | Arcee AI: Spotlight | Arcee AI | Multimodal | 62 | 131k ctx · $0.18/1M avg |
| 17 | NVIDIA: Nemotron Nano 12B 2 VL ≈7 GB VRAM (estimación baja de inferencia 4-bit) | NVIDIA | Multimodal | 62 | 131k ctx · $0.40/1M avg |
| 18 | ByteDance: UI-TARS 7B ≈4 GB VRAM (estimación baja de inferencia 4-bit) | ByteDance | Multimodal | 62 | 128k ctx · $0.15/1M avg |
| 19 | Baidu: ERNIE 4.5 VL 424B A47B ≈234 GB VRAM (estimación baja de inferencia 4-bit) | Baidu | Multimodal | 62 | 123k ctx · $0.83/1M avg |
| 20 | Reka Edge | Reka | Multimodal | 62 | 16k ctx · $0.10/1M avg |