DeepSeek
Contexte1M
Entrée$0.3000/M
Sortie$1.2000/M
A strong high-throughput, lower-cost default for general and agentic work: it natively understands images and supports a 1M-token context, optional thinking and non-thinking modes, tool calls, and JSON output. Compared with the higher-tier V4 Pro, it is positioned around speed, throughput, and cost efficiency, making it practical for long documents, coding/agents, and image-text analysis; enable thinking when deeper reasoning is needed.
Type d'entrée:
Type de sortie: