Falcon-H1 models
4 Falcon-H1 models from TII on Hugging Face, from 521M to 7.6B parameters, published by TII in BF16, with contexts from 16K to 256K tokens. The smallest official model, Falcon-H1-0.5B-Base, needs about 1.2 GB of VRAM at its published precision; the cheapest live fit is RTX 3070 at $0.088/hr.
Part of the Falcon series
Pick a size
One row per official Falcon-H1 size: the VRAM it needs at each precision, the cheapest GPU that holds it today, and the KV cache for a 32K-token context.
| Model | Parameters | Native VRAM | FP8 VRAM | INT4 VRAM | Live GPU fit (native) | Est. $/hr | KV cache at 32K |
|---|---|---|---|---|---|---|---|
| Falcon-H1-0.5B-Base | 521M | 1.2 GB | 0.6 GB | 0.3 GB | RTX 3070 | $0.088/hr | 0.56 GB |
| Falcon-H1-1.5B-Deep-Instruct | 1.6B | 3.5 GB | 1.7 GB | 0.9 GB | RTX 3070 | $0.088/hr | 2.1 GB |
| Falcon-H1-7B-Instruct | 7.6B | 17.0 GB | 8.5 GB | 4.2 GB | RTX A5000 | $0.176/hr | 1.4 GB |
VRAM is weights times a flat 1.2 overhead; the KV cache is a separate per-model figure at 16-bit, shown where the architecture is published. See the methodology.
Official models (4)
| Model | Parameters | Published as | VRAM needed | Live GPU fit | Est. $/hr |
|---|---|---|---|---|---|
| Falcon-H1-0.5B-Base | 521M | BF16 | 1.2 GB | RTX 3070 | $0.088/hr |
| Falcon-H1-0.5B-Instruct | 521M | BF16 | 1.2 GB | RTX 3070 | $0.088/hr |
| Falcon-H1-1.5B-Deep-Instruct | 1.6B | BF16 | 3.5 GB | RTX 3070 | $0.088/hr |
| Falcon-H1-7B-Instruct | 7.6B | BF16 | 17.0 GB | RTX A5000 | $0.176/hr |
VRAM is for the precision the model is published in: the weights times a flat 1.2 overhead, with the KV cache not included. See the methodology. The fit shown is the lowest-priced single GPU type that holds the model at that precision, or the lowest-priced multi-GPU set (up to 8) when none does.
More from TII
- Falcon 1 (4 models)
- Falcon Mamba (2 models)
- Falcon3 (4 models)