deepseek-v4.1POST

DeepSeek V4.1 Flash on one DGX Spark: 35.5 tok/s with 1.62bpw EXL3 quant

A SAGE-EXL3 1.62bpw quant plus vLLM recipe runs V4.1 Flash on a single DGX Spark at 35.5 tok/s decode and 1,189 tok/s prefill, with 99.61% HumanEval answer-token agreement.

Neo

@NeoAIForecastArchived 11 days agoOriginally posted 2026.10.20

Neo @NeoAIForecastOriginally posted 2026.10.20

Open original on X

One DGX Spark. DeepSeek V4.1 Flash 1017. 1.62 bits per expert. @ViC305 just dropped a SAGE-EXL3 quant and a one-Spark vLLM recipe!!! >35.5 tok/s decode >1,189 tok/s max prefill >229K context so far >99.61% answer-token agreement on HumanEval vs the native checkpoint weights = https://huggingface.co/vcruz305/DSV4.1-Flash-1017-SAGE-EXL3-1.62bpw recipe = https://github.com/vcruz305/DSV4.1-Flash-1017-EXL3-Spark-recipe

x.com/neoaiforecast/status/2112386229564285133Archived 2026.10.20 12:02