The free AI already on your Mac. CLI tool, OpenAI-compatible server, and interactive chat — all on-device via Apple Intelligence. No API keys, no cloud, no downloads.
- apple-intelligence
- apple-silicon
- cli
- foundationmodels
140,194 visible repositories in the index
1–11 of 11 · Most stars
The free AI already on your Mac. CLI tool, OpenAI-compatible server, and interactive chat — all on-device via Apple Intelligence. No API keys, no cloud, no downloads.
Rapid-MLX is an open-source (Apache 2.0) OpenAI- and Anthropic-compatible LLM inference server for Apple Silicon, built on MLX, focused on reliable tool calling for coding agents. Up to 4× faster than Apple's MLX (mlx-lm) on the same weights.
The fastest way to run Qwen 3.8 Flash Next, Qwen 3.8 27B and Ternary Bonsai 2 27B on a Mac: 125 tok/s in OpenCode on an M5 Max, and a 27B model on 16 GB Macs. Native MTP speculative decoding on Apple Silicon, exact at any temperature. OpenAI and Anthropic compatible local server.
OpenAI-compatible proxy for ChatGPT Codex Responses API
易创:纯本地、开源的 AI 网文写作桌面软件,支持小说创作、AI 辅助写作、BYOK 与自定义提示词。
本地优先的多模型 AI 图片生成工作台,支持 GPT Image、Gemini、图像编辑、素材库、模板与并发任务。Local-first AI image workbench for GPT Image & Gemini, with editing, reusable assets, templates and task history.
TranslatedFerry OpenCode's free models over to WorkBuddy — isolated runtime, concurrent detection, verdict caching, one-click import, and a local OpenAI-compatible forwarding port
Capri — an AI desktop companion for Windows with local/API models, Hugging Face, optional voice, memory and permission-based desktop tools.
Fast local LLM inference for the GPU you already have: one Rust binary, GGUF models, an OpenAI-compatible server and a chat page with a ghost.
PROMETHEVS: run the full DeepSeek-V4.1-Flash-1017 (510 GB) on a gaming PC with 64 GB RAM and a 12 GB GPU. PROMETHEVS engine (compress-then-correct), OpenAI-compatible local server, report and logs.
Run Qwen3.8-27B locally on Windows with one NVIDIA RTX GPU: chat and an OpenAI-compatible API. On an RTX 5090 the fastest model averaged 2,298 tok/s on our own JSONL edit task (one request at a time, thinking off, v0.1.4).