Qwen3.8-27B

FP8 · sglang · 128K context · Instinct MI300X 192 GB

Alibaba's 27B dense model from August 2026, and the strongest model that fits one consumer card. It thinks before it answers, calls tools reliably and reads images, which makes it the default for coding agents and long projects.

– tok/sdecode

– tok/sprefill

128Kcontext window

Earlier checkearlier

Run it

With Omarchy Local AI it is one button. By hand, it is three steps: the weights, the config, the container. The server then answers on http://localhost:8000/v1.

1 Download the weights

hf download Qwen/Qwen3.8-27B-FP8 \
  --revision 017b9c7af6b5689d5dd426a76e0bc077eb5ca20a

2 Start the server

docker run --rm \
  --device /dev/kfd --device /dev/dri \
  -p 8000:30000 \
  --shm-size 16g \
  -e HF_HOME=/root/.cache/huggingface \
  -v ~/.cache/huggingface:/root/.cache/huggingface \
  --entrypoint /opt/venv/bin/python \
  lmsysorg/sglang-rocm:v0.5.18-rocm720-mi30x-20260823@sha256:8805fe3744f9b549b4e329b37134bce6ce4afd7ec749ea1249507bcbcea10096 \
  -m sglang.launch_server \
  --model-path Qwen/Qwen3.8-27B-FP8 \
  --revision 017b9c7af6b5689d5dd426a76e0bc077eb5ca20a \
  --served-model-name qwen3.8-27b \
  --trust-remote-code \
  --host 0.0.0.0 \
  --port 30000 \
  --context-length 262144 \
  --mem-fraction-static 0.90 \
  --max-running-requests 4 \
  --attention-backend aiter \
  --mamba-ssm-dtype bfloat16 \
  --mamba-radix-cache-strategy extra_buffer \
  --chunked-prefill-size 2048 \
  --max-prefill-tokens 16384 \
  --cuda-graph-max-bs-decode 4 \
  --enable-cache-report \
  --reasoning-parser qwen3 \
  --tool-call-parser qwen3_coder

What it passed

✓ loadLoads and serves within an hour

✓ chatAnswers a plain question and stops on its own

– reasoningThinks separately and gets 17 × 23 right

– toolsCalls a tool with the right arguments and uses the result

– contextRecalls a code buried in a prompt that fills 85% of the window

– speedDecodes at 15 tok/s or more

Passed the older check (loads and chats); a full six-check run is pending.

Details

Weights: Qwen/Qwen3.8-27B-FP8 @ 017b9c7af6 ›
Image: lmsysorg/sglang-rocm:v0.5.18-rocm720-mi30x-20260823@sha256:8805fe3744f9b549b4e329b37134bce6ce4afd7ec749ea1249507bcbcea10096
Engine profile: sglang-qwen3.8-27b-fp8-128k ›
Recipe file: registry/recipes/amd/mi300x-192gb/qwen3.8-27b.sglang.128k.json ›
Model released: Aug 5, 2026
Tested: earlier acceptance