DeepSeek: deepseek-chat API

deepseek-chat is a dynamic model alias for DeepSeek's official API. As of January 2026, it points to the DeepSeek-V3.2 flagship. Utilizing a 671B total parameter MoE architecture with 37B active parameters, it serves as a versatile powerhouse for general-purpose tasks. Unlike the reasoning-focused deepseek-reasoner, this model defaults to Non-Thinking mode to ensure ultra-low latency and high throughput. Powered by DeepSeek Sparse Attention (DSA), it supports an expanded 1.01 million token context window. It remains one of the world's most cost-effective production-grade APIs, excelling in multilingual conversation, creative writing, and high-concurrency applications.

Frequently Asked Questions

What is the context window of deepseek-chat?

deepseek-chat supports a context window of up to 128,000 tokens.

Does deepseek-chat support function calling?

Yes. deepseek-chat supports tool / function calling.

What is the knowledge cutoff of deepseek-chat?

The knowledge cutoff of deepseek-chat is 2024-06-30.