AnalysisDevelopersNovember 5, 2026

Delta Inference Engine v0.93 runs DeepSeek-V4.1-Flash-1017 on one RTX 3090

Read original source →github.com

Update keeps 41 tok/s at 192GB/IQ2 quantization while improving output quality, and now runs from 64 GB RAM with first token ~40% sooner. Adds Unsloth quants at full speed, EXL3, basic multi-GPU and public CI.

1 source

More stories today

Open the live feed