Loading…
Loading…
Text generation · served with ONNX Runtime GenAI · ↓ 17 downloads a month · View on Hugging Face
The same model packaged differently: a quantized version needs less GPU memory and runs on a cheaper machine.
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|---|---|---|---|
| OriginalQwen/Qwen3-32B | vLLM | 66 GB | 77.4 GB | $0.89/h 2x Nvidia A40 |
| AWQQwen/Qwen3-32B-AWQ | vLLM | 19 GB | 26.1 GB | $0.38/h 2x RTX 5060 Ti |
| FP8Qwen/Qwen3-32B-FP8 | vLLM | 34 GB | 42.7 GB | $0.46/h RTX A6000 |
| NVFP4 (RedHatAI)RedHatAI/Qwen3-32B-NVFP4 | vLLM | 21 GB | 27.6 GB | $0.38/h 2x RTX 5060 Ti |
| unsloth bnb 4bit (unsloth)unsloth/Qwen3-32B-unsloth-bnb-4bit | vLLM | 39 GB | 48.3 GB | $0.75/h 4x RTX 5060 Ti |
| NVFP4 (nvidia)nvidia/Qwen3-32B-NVFP4 | vLLM | 21 GB | 27.6 GB | $0.38/h 2x RTX 5060 Ti |
| quantized.w4a16 (RedHatAI)RedHatAI/Qwen3-32B-quantized.w4a16 | vLLM | 19 GB | 25.9 GB | $0.38/h 2x RTX 5060 Ti |
| Instruct DirectML INT4 (wekkel)wekkel/Qwen3-32B-Instruct-DirectML-INT4 | ONNX Runtime GenAI | 20 GB | 26.2 GB | $0.18/h Nvidia V100 |
Deploying ONNX models from the catalog is coming soon. Today, download the files from Hugging Face and upload them in Inference.
Prices are per hour for the whole machine and its disk, right now. Speed is Vast's DLPerf score of the machine, relative to the cheapest. Best value: the most speed per dollar. Only machines with a Vast reliability above 95% that download at 400 Mbit/s or more are offered: the machine fetches the model itself.