Techyon
Runs on CPU284B model in 64 GB RAM · per Synchopate

Issue 149 Nov

llama-modchip runs a large DeepSeek model without a GPU

llama-modchip lets DeepSeek-V4-Flash, a 284B-parameter model, run entirely on a CPU with 64 GB of RAM, Synchopate reports. The released toolkit includes llama.cpp patches, model-compression tools and runtime utilities, with code under Apache-2.0.

Sources

Codesynchopate/llama-modchip: 🐋 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.

Written by AI from the sources