Run Claude Code 100% on-device with local AI on Apple Silicon. MLX-native Anthropic-API server, 65 tok/s Qwen 3.5 122B, Llama 3.3 70B, Gemma 4 31B. Private, offline, airgap-ready. Built for NDA / legal / healthcare workflows.

Plug-and-play local AI studio: uncensored chat, image & video generation, coding agent. Runs abliterated LLMs + ComfyUI 100% offline. One installer, no Docker, no cloud.

Docker Compose for llama.cpp GGUF servers on AMD Strix Halo: Qwen, Gemma, and Laguna packages (abliterated and quantized), stock Vulkan plus ROCmFP4/MTP and ROCmFPX, parallel slots, with prefill/decode and quality metrics measured on this rig.

Gemma 4 31B Abliterated — quality-preserving guardrail removal for Google's most capable open model. Apache 2.0. Runs on Apple Silicon via MLX.

Reproducible recipe: serve abliterated Gemma-4-12B (gemma4_unified) at 50-118 tok/s on no-NVLink Blackwell (SM120) via vLLM nightly + ModelOpt FP8/NVFP4 + MTP spec-decode.

Browser-based Claude-Code-style agent harness for local uncensored Ollama models. MoE-fast, security-audit aware, MIT.

Serving recipe for the abliterated (refusal-suppressed) GLM-5.2 on 4x DGX Spark: vLLM TP4, 200K context, 32 tok/s — with a full troubleshooting guide (model-id 404, worker wedge, decode stall, ops traps).

Abliterated GLM-5.2 with a true 4-bit NVFP4 KV cache on 4x DGX Spark: 316K context, 317,279-token pool (+58.6% vs fp8), 41.4 tok/s peak — with honest content-dependent decode measurements.

Anemll + MiaAI-Lab dual DGX Spark DSV4F+DSpark on vLLM 0.25.2.dev0 — 100% abliterated weights, Spec-Kit one-shot agent package. Full credit Anemll/MiaAI.

Reproducible vLLM recipe for shawnw3i/Huihui-Qwen3.6-27B-abliterated-AWQ-MTP on 2× RTX 3090 in a Proxmox LXC. MTP n=3, 256K context, full vision+tool-calling+reasoning. Silent 24/7 operation at 250W per card. Companion to the base-model recipe.

Run local coding agents with full tool access and browser-based monitoring using an uncensored Ollama model on your hardware.

Serve an abliterated Gemma-4-12B at high speeds on Blackwell GPUs without NVLink using vLLM, FP8 quantization, and MTP speculative decoding.

Run Gemma 4 31B locally on Apple Silicon with full guardrail removal and no quality loss.

Simple wrapper to run Strix (usestrix/strix) against a local OMLX MLX server with an abliterated/Heretic'd uncensored model on Apple Silicon.