Granite 4.2 30B
NVFP4 · vllm · 128K context · RTX PRO 6000 Blackwell · 96 GB
IBM's 30B dense model (64 layers). It thinks before it answers and calls tools.
79 tok/sdecode
3,843 tok/sprefill
128Kcontext window
Tested on this cardOct 30, 2026
Follow the steps below to download and run this recipe. Omarchy Local AI lists the recipes it can launch on your hardware. The server then answers on http://localhost:8000/v1.
- 1 Download the weights
hf download ibm-granite/granite-4.2-30b-nvfp4 \ --revision bddf34bc6db17e5c4f586e49c1668c8e97610446 \ --local-dir ~/models/granite-4.2-30b-nvfp4-bddf34bc
- 2 Start the server
docker run --rm \ --gpus all \ -p 8000:8000 \ --shm-size 16g \ -v ~/models/granite-4.2-30b-nvfp4-bddf34bc:/models/granite-4.2-30b-nvfp4-bddf34bc:ro \ --entrypoint vllm \ vllm/vllm-openai:v0.30.0@sha256:8a69ffad015f138d7170c4ddc429e230a3bc1c1719f67e14324749df200a4b90 \ serve \ /models/granite-4.2-30b-nvfp4-bddf34bc \ --served-model-name granite-4.2-30b-nvfp4-bddf34bc \ --host 0.0.0.0 \ --port 8000 \ --max-model-len 131072 \ --max-num-seqs 1 \ --gpu-memory-utilization 0.9 \ --kv-cache-dtype fp8 \ --enable-auto-tool-choice \ --tool-call-parser qwen3_coder \ --reasoning-parser deepseek_r1
✓ 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 all six checks on a real RTX PRO 6000 Blackwell WS (450 W cap) on Oct 30, 2026.
- Weights
- ibm-granite/granite-4.2-30b-nvfp4 @ bddf34bc6d ›
- Image
- vllm/vllm-openai:v0.30.0@sha256:8a69ffad015f138d7170c4ddc429e230a3bc1c1719f67e14324749df200a4b90
- Engine profile
- vllm ›
- Recipe file
- registry/recipes/nvidia/rtx-pro-6000-blackwell-96gb/granite-4.2-30b.vllm.128k.json ›
- Model released
- Aug 7, 2026
- Tested
- owner, RTX PRO 6000 Blackwell WS (450 W cap)