argonautlabsai/argodrive
Layout, balancer and instruments for running mixture-of-experts models from SSDs. Four models, two engines.
14 best points · 3 comments · 3 Hacker News posts · 6 GitHub stars · View on GitHub
Hacker News discussions
- Deep Seek v4.1 M5 Max at 17 tokens/s (14 points, 2 comments, by Argonautlabs, 2026-09-16)
- Show HN: GLM-5.3 744B at 4 tok/s on a MacBook Pro, experts streamed from 4 SSDs (2 points, 0 comments, by Argonautlabs, 2026-09-17)
- DeepSeek v4.1 Flash (518GB, 4-bit) on a 128GB MacBook: 2.7x prefill, 17 tok/s (1 points, 1 comments, by Argonautlabs, 2026-09-15)