Comparar modelos
Compare modelos lado a lado por preço, contexto, capacidades e benchmarks.
DeepSeek V3.2 is DeepSeek's V3-generation model upgrade underlying both the deepseek-chat (non-thinking) and deepseek-reasoner (thinking) modes, improving efficiency and reasoning quality over V3.1. It is designed as a strong, low-cost general-purpose and reasoning model.
Claude Opus 5.5 is Anthropic's flagship model for demanding reasoning, coding, and long-horizon agentic work, succeeding Claude Opus 5.
Benchmarks
Pontuações frente a frente nos índices de raciocínio, programação e agentes, além de classificações por domínio.