Xolotl runs DeepSeek's 552B-parameter model on a single gaming PC
AI-generated summary Tokenstrom reports that the open-source project Xolotl runs DeepSeek's 552B-parameter model on a single gaming PC. The software uses expert tiering and custom quantization to spread sub-networks across video memory, system RAM and an SSD.
- Lawyer upgrades eight-V100 private AI server to run 80% of DeepSeek V4.1 Flash at 4-bit NVFP4 precision — adds 1TB of RAM, an Nvidia RTX 3090, and fixes crashes caused by a bad PSU
- Lawyer builds private AI server with eight used Nvidia V100 GPUs — up to 215 tokens per second to keep legal files offline, costing about $2,400 per four-GPU board