---
type: "Article"
title: "DeepSeek-V4.1-Flash EXL3 2.0 bpw with vLLM on 2× DGX Spark"
description: "DeepSeek-V4.1-Flash EXL3 2.0 bpw with vLLM on 2× DGX Spark: recipe by sfxnz with speed, condition and source for every number, weights and license."
resource: "https://www.contextstudios.ai/local-ai/sfxnz--dsv41-flash-exl3-2.0bpw-viterbi-2x-dgx-spark"
language: "en"
generated:
  by: "process:contextstudios-md/1"
  at: "2026-10-02T20:37:01.106Z"
status: "stable"
---

# DeepSeek-V4.1-Flash EXL3 2.0 bpw with vLLM on 2× DGX Spark

DeepSeek-V4.1-Flash EXL3 2.0 bpw with vLLM on 2× DGX Spark: 42.6 tok/s according to github.com (dataset as of Sep 29, 2026).

- Creator: [sfxnz](https://www.contextstudios.ai/local-ai/creators/sfxnz.md)
- Engine: vLLM
- Quantization: EXL3 2.0 bpw MCG (routed experts, tail-biting Viterbi re-encoded, scale refit), lm_head MXFP8, KV fp8 (8-GiB-Pin), DSpark-3
- Model family: DeepSeek-V4.1-Flash
- Hardware: dgx-spark x2
- Context: 1048576
- Artificial Analysis Intelligence Index: 39.5 ([source](https://artificialanalysis.ai/models/deepseek-v4-1-flash))

## All numbers

- 42.6 tok/s (Everyday; Everyday, 1 request, prose prompt, with DSpark (per recipe)) — [source](https://github.com/sfxnz/DeepSeek-V4.1-Flash-EXL3-vLLM-2x-DGX-Spark/blob/main/README.md)
- decode_tps: 42.58 (1 request) — [source](https://github.com/sfxnz/DeepSeek-V4.1-Flash-EXL3-vLLM-2x-DGX-Spark/blob/main/README.md)
- decode_tps: 84.2 (1 request) — [source](https://github.com/sfxnz/DeepSeek-V4.1-Flash-EXL3-vLLM-2x-DGX-Spark/blob/main/README.md)
- prefill_tps: 838 (1 request) — [source](https://github.com/sfxnz/DeepSeek-V4.1-Flash-EXL3-vLLM-2x-DGX-Spark/blob/main/README.md)

## Weights

- [sfxnz/DeepSeek-V4.1-Flash-EXL3](https://huggingface.co/sfxnz/DeepSeek-V4.1-Flash-EXL3) (main, mit)
- [deepseek-ai/DeepSeek-V4.1-Flash](https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash) (base, mit)

[https://github.com/sfxnz/DeepSeek-V4.1-Flash-EXL3-vLLM-2x-DGX-Spark](https://github.com/sfxnz/DeepSeek-V4.1-Flash-EXL3-vLLM-2x-DGX-Spark)

## Related

- [Local AI on your hardware.](https://www.contextstudios.ai/local-ai.md)