signalnine/q27

Qwen3.8-27b and Qwen3.6-35B-A3B local inference engine for Vulkan and Metal

Blazing fast Apple Silicon optimized inference engine built for Qwen and Nemotron. OpenAI + Anthropic API compatible. Built by folks from nace.ai

Fork of antirez/ds4 for Apple Silicon, tuned mostly for M1/M2: Qwen3.8-Flash-Next with 262K context on 64 GB and up, 35 tok/s on an M1 Max, MTP and vision, one-line install.

A custom inference engine tuned for Qwen4 architecture. Currently runs Qwen3.8-Flash-Next at 15+ tok/s on 16GB of VRAM and 30 GB of RAM

Perplexity open source garden for inference technology

Run GreenBitAI's Quantized LLMs on Apple Devices with MLX

Fastest LLM inference runtime for Apple Silicon

Metal-first MoE inference for Apple Silicon with bounded SSD expert streaming and local OpenAI Chat, Responses, and Anthropic Messages endpoints.

works well with rocks; native Apple Silicon inference engine for capable and tiny models (ternary, Qwen) via MLX Swift. Hadamard-rotated KV cache, vision, tool calling, OpenAI + Anthropic APIs