Loading…
Loading…
Models / Falcon-H1
5 models (0.5B to 7B) · ↓ 113.1K downloads a month · from $0.10/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 |
|---|---|---|---|---|
| Instructtiiuae/Falcon-H1-1.5B-Deep-Instruct | vLLM | 3.1 GB | 6.3 GB | $0.10/h RTX 3060 |
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|---|---|---|---|
| Originaltiiuae/Falcon-H1-1.5B-Base | vLLM | 3.1 GB | 5.9 GB | $0.10/h RTX 3060 |
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|---|---|---|---|
| Instructtiiuae/Falcon-H1-7B-Instruct | vLLM | 15 GB | 19.5 GB | $0.19/h 2x RTX 3060 |
| Falcon H1R 7Btiiuae/Falcon-H1R-7B | vLLM | 15 GB | 19.5 GB | $0.19/h 2x 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").