Native inference engine optimized for DeepSeek V4 Flash, GLM 5.2, and PRO models with Metal, CUDA, and ROCm support
Terminal AI coding assistant optimized for DeepSeek v4 with deep thinking and agent skills
Agentic coding terminal powered by DeepSeek V4 with multi-provider LLM support.
High-performance C++ LLM inference engine — run DeepSeek 671B on a single GPU
A terminal-native coding agent powered by DeepSeek models with 1M-token context and thinking-mode reasoning
671B parameter MoE language model with state-of-the-art performance and efficient inference
