---
type: "Article"
title: "MiniMax-M3.1-Flash EXL3 2.15 bpw with vLLM on 1× DGX Spark"
description: "MiniMax-M3.1-Flash EXL3 2.15 bpw with vLLM on 1× DGX Spark: recipe by vcruz305 with speed, condition and source for every number, weights and license."
resource: "https://www.contextstudios.ai/local-ai/vcruz305--minimax-m31-flash-exl3-2.15-1spark-vllm"
language: "en"
generated:
  by: "process:contextstudios-md/1"
  at: "2026-11-11T03:30:15.906Z"
status: "stable"
---

# MiniMax-M3.1-Flash EXL3 2.15 bpw with vLLM on 1× DGX Spark

MiniMax-M3.1-Flash EXL3 2.15 bpw with vLLM on 1× DGX Spark: 24.4 tok/s according to github.com (dataset as of Nov 3, 2026).

- Creator: [vcruz305](https://www.contextstudios.ai/local-ai/creators/vcruz305.md)
- Engine: vLLM (vllm/vllm-openai:minimax-m31-flash, Digest 86f582da) + vllm-exl3 Fork 41fdb7d + exllamav3 Fork fdd7736, DSpark k=4
- Quantization: EXL3 SAGE 2.15bpw, q4-KV (nvfp4-Layout), fp8 Indexer-/Draft-KV
- Model family: MiniMax-M3.1-Flash
- Hardware: dgx-spark x1
- Context: 49152

## All numbers

- 24.4 tok/s (Everyday; Everyday, 1 request, prose prompt, with DSpark (per recipe)) — [source](https://github.com/vcruz305/MiniMax-M3.1-Flash-EXL3-Spark-recipe#measured-results)
- decode_tps: 24.37 (1 request) — [source](https://github.com/vcruz305/MiniMax-M3.1-Flash-EXL3-Spark-recipe#measured-results)
- decode_tps: 29.72 (1 request) — [source](https://github.com/vcruz305/MiniMax-M3.1-Flash-EXL3-Spark-recipe#measured-results)
- prefill_tps: 379.4 (1 request) — [source](https://github.com/vcruz305/MiniMax-M3.1-Flash-EXL3-Spark-recipe#measured-results)
- ttft_ms: 17500 (1 request) — [source](https://github.com/vcruz305/MiniMax-M3.1-Flash-EXL3-Spark-recipe#measured-results)

## Weights

- [vcruz305/MiniMax-M3.1-Flash-EXL3-SAGE-2.15bpw](https://huggingface.co/vcruz305/MiniMax-M3.1-Flash-EXL3-SAGE-2.15bpw) (main, other)
- [MiniMaxAI/MiniMax-M3.1-Flash](https://huggingface.co/MiniMaxAI/MiniMax-M3.1-Flash) (base, other)

[https://github.com/vcruz305/MiniMax-M3.1-Flash-EXL3-Spark-recipe](https://github.com/vcruz305/MiniMax-M3.1-Flash-EXL3-Spark-recipe)

## Related

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