Models / falcon
falcon by tiiuae
6 models (7B to 180B) · ↓ 284.7K downloads a month · from $0.24/h on RTX 3090 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.
falcon 7B
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|
| Originaltiiuae/falcon-7b | vLLM | 14 GB | 23.6 GB | $0.24/h RTX 3090 |
falcon 7B
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|
| Originaltiiuae/falcon-7b-instruct | vLLM | 14 GB | 23.6 GB | $0.24/h RTX 3090 |
falcon 11B
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|
| Originaltiiuae/falcon-11B | vLLM | 22 GB | 35.8 GB | $0.45/h Nvidia A40 |
falcon 40B
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|
| Originaltiiuae/falcon-40b | vLLM | 84 GB | 113.1 GB | $1.38/h 8x RTX 3090 |
falcon 180B
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|
| Originaltiiuae/falcon-180B-chatLicence to accept | vLLM | 359 GB | ~433.1 GB | $11.77/h 8x RTX PRO 6000 |
falcon 180B
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|
| Originaltiiuae/falcon-180BLicence to accept | vLLM | 359 GB | ~433.1 GB | $11.77/h 8x RTX PRO 6000 |
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").