Loading…
Loading…
Models / sarashina2.2
2 models (0.5B to 3B) · ↓ 31.3K downloads a month · from $0.07/h on GTX 1080 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 |
|---|---|---|---|---|
| instruct v0.1sbintuitions/sarashina2.2-0.5b-instruct-v0.1 | vLLM | 1.6 GB | 4.5 GB | $0.09/h RTX 3060 |
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|---|---|---|---|
| instruct v0.1sbintuitions/sarashina2.2-3b-instruct-v0.1 | vLLM | 6.7 GB | 11.2 GB | $0.09/h RTX 3060 |
| instruct ryzenai npu (ryugyosoft)ryugyosoft/sarashina2.2-3b-instruct-ryzenai-npu | ONNX Runtime GenAI | 540 MB | 3.5 GB | $0.07/h GTX 1080 |
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").