DeepSeek-R1
MIT-licensed open-weight reasoning model trained via reinforcement learning for math, coding, and verifiable chain-of-thought tasks.
General chat assistants and the foundation models behind them.
MIT-licensed open-weight reasoning model trained via reinforcement learning for math, coding, and verifiable chain-of-thought tasks.
Open-weight Mixture-of-Experts foundation model engineered for high-efficiency general chat, coding, and bilingual comprehension.
Meta open-weight foundation model family widely supported across local inference engines, cloud providers, and enterprise stacks.
European frontier AI lab delivering fast open-weight and commercial models alongside Le Chat and enterprise deployment options.
Open-source local model runner making it simple to pull, serve, and run open-weight LLMs and vision models via CLI or OpenAI-compatible API.
Alibaba open-weight dense and MoE model family supporting hybrid thinking modes, agentic tool calling, and wide multilingual coverage.