2 tracked versions of the DeepSeek family. Timeline ordered newest → oldest. Flagship version marked.
DeepSeek R1 is an open-weight reasoning model in the o1-class trained with reinforcement learning. Exposes visible chain-of-thought with top-tier math and competitive-coding benchmarks. MIT licensed.
DeepSeek V3 is a 671B-parameter MoE language model with strong reasoning and coding performance at prices roughly 1/10 of frontier US models. Open-weights released under DeepSeek License v1.0; OpenAI-compatible official API available globally.
See every provider hosting DeepSeek AI models at /provider/deepseek.
Your brand here — Reach our audience of professional directory owners and boost your sales.