DeepSeek
Open-weight reasoning models with an OpenAI-compatible API
DeepSeek is profiled here as a LLM tool for engineering teams. Read about features, pricing, and how it compares to related options in the tools directory.
Description
DeepSeek is an AI research lab in Hangzhou, China, founded in 2023 by Liang Wenfeng and funded by the quantitative hedge fund High-Flyer. The lab releases open-weight language and reasoning models under permissive licenses and serves them through a token-priced, OpenAI-compatible API. Teams pick DeepSeek when they need to inspect, fine-tune, or self-host the weights behind their application. The R1 release in January 2025 demonstrated that open-weight reasoning models could approach closed frontier systems on math and coding benchmarks, and the lab has shipped successive V3 and R1 updates since.
Key Capabilities:
Open-weight V3 and R1 model families under MIT licensing
Reasoning models with exposed chain-of-thought traces
OpenAI-compatible chat completions API
Context caching that lowers the cost of repeated prompt prefixes
Function calling and JSON output modes
Self-hostable through vLLM, SGLang, and Ollama
Alternative tools
- Kong AI Gateway
AI traffic governance built on the Kong gateway
- Eden AI
One API for many AI models across providers and modalities
- Reka
Multimodal models that reason across text, image, and video
- Moonshot AI
Open-weight Kimi models tuned for agentic and coding tasks
- xAI Grok
Grok model family with real-time search through one API
- LLM Guard
Open-source security toolkit for LLM interactions