Loading…
Loading…
Models / Qwen3-VL-Embedding
2 models (2B to 8B) · ↓ 2.3M downloads a month · from $0.09/h on RTX 3060 right now. Each model's versions do the same job; quantized ones (AWQ, FP8, 4-bit) need less GPU memory and run on cheaper machines.
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|---|---|---|---|
| OriginalQwen/Qwen3-VL-Embedding-2B | vLLM | 4.3 GB | 6.6 GB | $0.09/h RTX 3060 |
| AWQ 4bit (LifetimeMistake)LifetimeMistake/Qwen3-VL-Embedding-2B-AWQ-4bit | vLLM | 2.8 GB | 4.8 GB | $0.09/h RTX 3060 |
| FP8 (shigureui)shigureui/Qwen3-VL-Embedding-2B-FP8 | vLLM | 2.9 GB | 4.9 GB | $0.09/h RTX 3060 |
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|---|---|---|---|
| OriginalQwen/Qwen3-VL-Embedding-8B | vLLM | 16 GB | 21 GB | $0.23/h RTX A5000 |
| FP8 (RamManavalan)RamManavalan/Qwen3-VL-Embedding-8B-FP8 | vLLM | 11 GB | 14.2 GB | $0.13/h RTX 4060 Ti |
| AWQ 4bit (gonuit)gonuit/Qwen3-VL-Embedding-8B-AWQ-4bit | vLLM | 6.0 GB | 8.7 GB | $0.09/h RTX 3060 |
Model details come from Hugging Face and are refreshed daily; prices follow the live offers. GPU memory is GridPup's estimate: the weights, room for an 8,192-token conversation (chat models) and the runtime's own memory. Large models are split across the GPUs of one machine ("4x RTX 4090").