---
type: "Article"
title: "Qwen3.8-Flash-Next FP8 with vLLM on 8× RTX PRO 6000"
description: "Qwen3.8-Flash-Next FP8 with vLLM on 8× RTX PRO 6000: recipe by helixml with speed, condition and source for every number, weights and license."
resource: "https://www.contextstudios.ai/local-ai/helixml--qwen38-flash-next-fp8-vllm-2xtp4-8xpro6000"
language: "en"
generated:
  by: "process:contextstudios-md/1"
  at: "2026-11-10T11:21:59.810Z"
status: "stable"
---

# Qwen3.8-Flash-Next FP8 with vLLM on 8× RTX PRO 6000

Qwen3.8-Flash-Next FP8 with vLLM on 8× RTX PRO 6000: 202 tok/s according to github.com (dataset as of Nov 3, 2026).

- Creator: [helixml](https://www.contextstudios.ai/local-ai/creators/helixml.md)
- Engine: vLLM (vllm/vllm-openai@sha256:0aea3024…, 2× TP4 + Expert-Parallel, MTP3) hinter ramjet v0.8.0
- Quantization: FP8 (offizieller Checkpoint Qwen/Qwen3.8-Flash-Next-FP8)
- Model family: Qwen3.8-Flash-Next
- Hardware: rtx-pro-6000 x8
- Context: 262144
- Artificial Analysis Intelligence Index: 39.8 ([source](https://artificialanalysis.ai/models/qwen3-8-flash-next))

## All numbers

- 202 tok/s (Peak; Peak, 1 request, best value reported by source, with MTP3) — [source](https://github.com/helixml/ramjet#models-with-a-validated-stack)
- decode_tps: 202 (1 request) — [source](https://github.com/helixml/ramjet#models-with-a-validated-stack)

## Weights

- [Qwen/Qwen3.8-Flash-Next-FP8](https://huggingface.co/Qwen/Qwen3.8-Flash-Next-FP8) (main, other)
- [Qwen/Qwen3.8-Flash-Next](https://huggingface.co/Qwen/Qwen3.8-Flash-Next) (base, other)

[https://github.com/helixml/ramjet](https://github.com/helixml/ramjet)

## Related

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