Benchmarks de modelos de IA

O Reviuws executa uma suíte de testes padrão de benchmark abarcando raciocínio, programação, latência e eficiência de custo, e registra uma pontuação de desempenho para cada modelo no diretório.

• GPT-5.5 benchmark results
• SDXL benchmark results
• Runway Gen-4 benchmark results
• Whisper Large v3 benchmark results
• text-embedding-3-small benchmark results
• GPT Image 2 benchmark results
• o4-mini benchmark results
• Gemini 2.5 Pro benchmark results
• Gemini 2.5 Flash-Lite benchmark results
• text-embedding-004 benchmark results
• Claude Opus 4.1 benchmark results
• Claude Haiku 4.5 benchmark results
• Mistral Large 2 benchmark results
• Codestral 2508 benchmark results
• Whisper-1 benchmark results
• Gemini 2.5 Flash benchmark results
• Gemma 3 benchmark results
• Claude Sonnet 4.5 benchmark results
• Llama 3.3 70B benchmark results
• Mistral Small 3.2 benchmark results
• DeepSeek-V3.1 benchmark results
• DeepSeek-R1 benchmark results
• Qwen2.5-VL-72B benchmark results
• Grok 4.3 benchmark results
• Veo 3.1 benchmark results
• Qwen3-235B-A22B benchmark results
• Qwen 3 235B benchmark results
• Gemini Embedding 2 benchmark results
• Claude Sonnet 5 benchmark results
• GPT-5.6 Luna benchmark results
• Claude Sonnet 4.5 benchmark results
• Grok 4.5 benchmark results
• GPT-5.6 Terra benchmark results
• Claude Fable 5 benchmark results
• Claude Opus 5 benchmark results
• GPT-Realtime 2.1 benchmark results
• DeepSeek V4 Pro benchmark results
• Claude Haiku 4.5 benchmark results
• GPT-5.6 Sol benchmark results
• GPT-5.4 mini benchmark results