# Flash-Moe

11 Hacker News commenters recommended Flash-Moe and 3 criticized it, November 2025 to November 2026. It was recommended in 3 of 13 months.

**Verdict:** Open-source engine that runs huge mixture-of-experts models on a Mac by streaming weights from SSD; praised as clever engineering, but throughput is low for everyday use.

## Why people recommend it

- Runs huge models on a Mac (2 comments): "Flash-MoE is basically turning “impossible” into a weekend project." ([source](https://news.ycombinator.com/item?id=47487716)) "over 200B params with flash-moe occasionally" ([source](https://news.ycombinator.com/item?id=47873904))
- Clever engineering and write-up (2 comments): "The technical write up is great, but Mac users should not get too excited just yet on running 300B+ parameter models locally as the TPS isn't that good." ([source](https://news.ycombinator.com/item?id=47476725)) "This project shows an interesting automated search for engineering problems" ([source](https://news.ycombinator.com/item?id=47484838))
- Inspired other projects and adaptations (2 comments): "I'm sure people will adapt the Flash-MoE integration to be even better soon as well" ([source](https://news.ycombinator.com/item?id=47925279)) "Huge respect to them, it was a big inspiration for this project!" ([source](https://news.ycombinator.com/item?id=49099667))

## What people warn about

- Too slow for daily use (4 comments): "6t/s is unbearable, if you used it with OpenCode you would be waiting 20+ minutes per turn." ([source](https://news.ycombinator.com/item?id=47490480)) "they are doing similar things with bigger models, but tok/s is not high" ([source](https://news.ycombinator.com/item?id=49101028)) "it's not really usable for non-niche purposes due to the low throughput" ([source](https://news.ycombinator.com/item?id=47594605))

## Alternatives hackers recommend

- [vLLM](https://hackerapproved.com/p/vllm.md): 112 recommend
- [Strata](https://hackerapproved.com/p/strata.md): 31 recommend
- [DwarfStar 4](https://hackerapproved.com/p/dwarfstar.md): 26 recommend
- [SGLang](https://hackerapproved.com/p/sglang.md): 19 recommend
- [ninfer](https://hackerapproved.com/p/ninfer.md): 16 recommend

## Cite as

Hacker Approved, "Flash-Moe", https://hackerapproved.com/p/flash-moe/ (counts from 1,795,640 Hacker News comments, November 2025 to November 2026).