Loading…
Loading…
Models / zephyr
3 models (7B to 7B) · ↓ 96K downloads a month · from $0.09/h on Titan Xp 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 |
|---|---|---|---|---|
| OriginalHuggingFaceH4/zephyr-7b-beta | vLLM | 14 GB | 19.5 GB | $0.18/h 2x RTX 3060 |
| oga (SamTheDev)SamTheDev/zephyr-7b-beta-oga | ONNX Runtime GenAI | 4.3 GB | 7.5 GB | $0.09/h Titan Xp |
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|---|---|---|---|
| OriginalHuggingFaceH4/zephyr-7b-alpha | vLLM | 14 GB | 19.5 GB | $0.18/h 2x RTX 3060 |
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|---|---|---|---|
| zephyr 7b gemma v0.1HuggingFaceH4/zephyr-7b-gemma-v0.1 | vLLM | 17 GB | 25.4 GB | $0.38/h 2x RTX 5060 Ti |
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").