🐋 Run 284B DeepSeek-V4-Flash (<59GiB RAM, 44.4 tok/s pp, 9.1-14.7 tok/s) & 30B MoEs (25-111 tok/s) on a pure 64GB CPU at 256K ctx (0 GPUs). SwiGLU-H2, Stockfish-PVS L3 & 13ms Claude Code restore.
🐋 Run 284B DeepSeek-V4-Flash (<59GiB RAM, 44.4 tok/s pp, 9.1-14.7 tok/s) & 30B MoEs (25-111 tok/s) on a pure 64GB CPU at 256K ctx (0 GPUs). SwiGLU-H2, Stockfish-PVS L3 & 13ms Claude Code restore.
Quickstart · Quantization & Memory Specs · Supported Models · Supported Hardware · Leaderboard · Technical Paper (paper/) · Architecture · FAQ
llama-modchip is a set of ggml/llama.cpp kernel patches, quantization transcoders, and runtime tools designed to run sparse Mixture-of-Experts (MoE) checkpoints entirely in system RAM on x86_64 (AVX-512 / AVX2) and ARM64 (NEON / SVE / Apple Metal) processors:
Description and README excerpt from synchopate/llama-modchip on GitHub ↗.
synchopate/llama-modchip was first seen on GitHub · Rising today on Nov 8, 2026. Since then it has appeared in 2 checks on 1 different day. Its best position was #6 of 30, reached on Nov 8, 20:45 UTC. It dropped off the list after Nov 8, 21:00 UTC. The same link also trended on Hacker News, which raises its cross-platform score.
Position on GitHub · Rising today at every check (lower is better).
Best rank per day (UTC).
Platform metric as reported by GitHub.
| List | Best | Days | Checks | Last seen |
|---|---|---|---|---|
| GitHub · Rising today | #6 | 1 | 2 | |
| YHacker News · Front page Llama-modchip: 284B DeepSeek on 64GB RAM, no GPU | same link | |||
Archived daily, weekly and monthly lists that include this entry.
Not at the moment. It was last on the list on Nov 8, 21:00 UTC, after 1 day of trending.
#6 on GitHub · Rising today, on Nov 8, 20:45 UTC. Trendoday checks this list every 60 minutes.
It was first seen on Nov 8, 2026 and has been on the list on 1 day (2 checks) since.
Mostly Python; it is licensed under Apache-2.0. It has 418 stars and 25 forks.