GLM-4.5-Air-nvfp4

Format: NVFP4 — weights & activations quantized to FP4 with dual scaling.
Base model: zai-org/GLM-4.5-Air
How it was made: One-shot calibration with LLM Compressor (NVFP4 recipe), long-seq calibration with Rombo-Org/Optimized_Reasoning.

Notes: Keep lm_head in high precision; calibrate on long, domain-relevant sequences.

Check the original model card for information about this model.

Running the model with VLLM in Docker

sudo docker run --runtime nvidia --gpus all -p 8000:8000 --ipc=host -e VLLM_USE_FLASHINFER_MOE_FP4=1 vllm/vllm-openai:nightly --model Firworks/GLM-4.5-Air-nvfp4 --dtype auto --max-model-len 32768

Running the model on the DGX Spark with VLLM in Docker

sudo docker run --gpus all --network host --ipc=host   -e HF_TOKEN=$HF_TOKEN   -v ~/.cache/huggingface:/root/.cache/huggingface   nvcr.io/nvidia/vllm:26.02-py3   vllm serve Firworks/GLM-4.5-Air-nvfp4   --dtype auto   --max-model-len 32768   --enforce-eager

This was tested on a B200 cloud instance.

If there are other models you're interested in seeing quantized to NVFP4 for use on the DGX Spark, or other modern Blackwell (or newer) cards let me know. I'm trying to make more NVFP4 models available to allow more people to try them out.