this recipe is to be used with https://github.com/eugr/spark-vllm-docker

qwen3.6-35b-fp8.yaml

# Recipe: Qwen3.6-35B-A3B-FP8
# Qwen3.6-35B model in native FP8 quantization

recipe_version: "1"
name: Qwen3.6-35B-FP8
description: vLLM serving Qwen3.6-35B-FP8

# HuggingFace model to download (optional, for --download-model)
model: Qwen/Qwen3.6-35B-A3B-FP8

# Only cluster is supported
cluster_only: false

# Container image to use
container: vllm-node

# No mods required
mods: {}
#  - mods/fix-qwen3.5-chat-template

# Default settings (can be overridden via CLI)
defaults:
  port: 8000
  host: 0.0.0.0
  gpu_memory_utilization: 0.7
  max_model_len: 262144
  max_num_batched_tokens: 32768

# Environment variables
env:
  VLLM_MARLIN_USE_ATOMIC_ADD: '1'


# The vLLM serve command template
command: |
  vllm serve Qwen/Qwen3.6-35B-A3B-FP8 \
    --served-model-name Qwen3.6-35B-A3B \
    --max-model-len {max_model_len} \
    --gpu-memory-utilization {gpu_memory_utilization} \
    --port {port} \
    --host {host} \
    --load-format instanttensor \
    --enable-prefix-caching \
    --kv-cache-dtype fp8 \
    --attention-backend flashinfer \
    --enable-auto-tool-choice \
    --tool-call-parser qwen3_coder \
    --reasoning-parser qwen3 \
    --generation-config auto \
    --override-generation-config '{{"temperature": 1.0, "top_p": 0.95, "top_k": 20, "presence_penalty": 1.5, "repetition_penalty": 1.0}}' \
    --default-chat-template-kwargs '{{"preserve_thinking": true}}' \
    --speculative-config '{{"method":"mtp","num_speculative_tokens":3}}' \
    --max-num-batched-tokens {max_num_batched_tokens} \
    --max-num-seqs 8 \
    --stream-interval 5 \
    --max_cudagraph_capture_size 10 \
    --mamba_ssm_cache_dtype float16

works well for me, any reason why you are not using DFlash? https://huggingface.co/z-lab/Qwen3.6-35B-A3B-DFlash

@danielfl , it's just dropped less than day ago :) I'll try it this weekend.