Loading…
Loading…
Models / Qwen3-Embedding
3 models (0.6B to 8B) · ↓ 12.4M downloads a month · from $0.08/h on GTX 1660 SUPER 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-Embedding-0.6B | Text Embeddings Inference | 1.2 GB | 3 GB | $0.08/h GTX 1660 SUPER |
| ONNX (onnx-community)onnx-community/Qwen3-Embedding-0.6B-ONNX | ONNX Runtime | 570 MB | 1.9 GB | $0.08/h GTX 1660 SUPER |
| onnx (Qdrant)Qdrant/Qwen3-Embedding-0.6B-onnx | ONNX Runtime | 1.1 GB | 2.7 GB | $0.08/h GTX 1660 SUPER |
| ONNX (n24q02m)n24q02m/Qwen3-Embedding-0.6B-ONNX | ONNX Runtime | 540 MB | 1.8 GB | $0.08/h GTX 1660 SUPER |
| GraphQL (xthor)xthor/Qwen3-Embedding-0.6B-GraphQL | Text Embeddings Inference | 1.2 GB | 3 GB | $0.08/h GTX 1660 SUPER |
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|---|---|---|---|
| OriginalQwen/Qwen3-Embedding-4B | Text Embeddings Inference | 8.0 GB | 12 GB | $0.10/h RTX 3060 |
| W4A16 G128 (boboliu)boboliu/Qwen3-Embedding-4B-W4A16-G128 | Text Embeddings Inference | 2.6 GB | 4.9 GB | $0.08/h GTX 1660 SUPER |
| Octen Embedding 4B INT8 (Octen)Octen/Octen-Embedding-4B-INT8 | Text Embeddings Inference | 4.4 GB | 7.2 GB | $0.09/h RTX 3060 |
| FP8 Dynamic (DCC-BS)DCC-BS/Qwen3-Embedding-4B-FP8-Dynamic | Text Embeddings Inference | 4.4 GB | 7.2 GB | $0.09/h RTX 3060 |
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|---|---|---|---|
| OriginalQwen/Qwen3-Embedding-8B | Text Embeddings Inference | 15 GB | 21.2 GB | $0.24/h RTX 3090 |
| NVFP4 (alexliap)alexliap/Qwen3-Embedding-8B-NVFP4 | Text Embeddings Inference | 5.2 GB | 8.2 GB | $0.09/h RTX 3060 |
| FP8 DYNAMIC (alexliap)alexliap/Qwen3-Embedding-8B-FP8-DYNAMIC | Text Embeddings Inference | 8.2 GB | 12.1 GB | $0.14/h RTX A4000 |
| Octen Embedding 8B INT8 (Octen)Octen/Octen-Embedding-8B-INT8 | Text Embeddings Inference | 8.2 GB | 12.2 GB | $0.14/h RTX A4000 |
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").