carloslfu/slotstream

carloslfu/slotstream の 2026 年 10 月 25 日のコミットは Q2 検証の測定を記録し、直接予測読み込み候補を差し戻した。

GitHub の説明: Run a 105 GB AI model on a Mac that can't hold it. Slotstream streams Qwen3.8-Flash-Next (125B mixture of experts) from your SSD and caches the busiest experts in memory, so it runs on Macs with 16 to 64 GB. One native Swift binary on MLX and Metal, no Python, offline. Works with Claude Code, Codex and Ollama or OpenAI clients.