slotstream
Run a 105 GB AI model on a Mac that can't hold it. Qwen3.8-Flash-Next 4-bit, streamed from SSD on 16–64 GB Apple Silicon Macs. Fully native Swift, MLX and Metal
What is slotstream?
Run a 105 GB AI model on a Mac that can't hold it. Qwen3.8-Flash-Next 4-bit, streamed from SSD on 16–64 GB Apple Silicon Macs. Fully native Swift, MLX and Metal. Offline, no Python. CLI, Swift library, and Ollama/OpenAI/Anthropic APIs. Works with Claude Code and Codex.
Overview of slotstream
What it is
slotstream is a native Swift program for Apple Silicon Macs that runs the 125-billion-parameter Qwen3.8-Flash-Next mixture-of-experts model (105 GB in 4-bit) on machines with 16 to 64 GB of memory. It keeps most of the model on the SSD and streams the busiest experts into memory as needed, using Apple's MLX and Metal frameworks. No Python, no cloud account, and it works offline after a one-time download. It provides Ollama-, OpenAI- and Anthropic-compatible APIs so that external tools like Claude Code, Codex and others can connect to the local model.
Key features
- Runs a 105 GB model on Macs with 16–64 GB RAM by streaming experts from SSD - One native Swift binary using MLX and Metal, no Python dependency - Offline operation after initial download - Offers Fast and Smart model presets for different task difficulty - Provides Ollama-, OpenAI-, and Anthropic-compatible API endpoints - Includes a Homes data browser with search, filters, record details and relationships - Works with Claude Code, Codex, Pi, opencode and Hermes via the
slotstream launchcommand
Use cases
- Running frontier-class open models on Macs that cannot hold the entire model in memory - Local AI coding assistant via Claude Code, Codex or similar tools - Image understanding tasks (the model can process pictures)
- Developing custom apps that connect through standard API interfaces - Offline AI usage without internet or cloud accounts
Getting started
slotstream launch claude After installation, the slotstream launch command starts Claude Code using the local model. For full installation and setup details, see the repository README.
License and pricing
MIT license. Open source. Pricing is not mentioned beyond the license.
Based on the project's public README and repository metadata.