fishaudio models
2 fishaudio models from fishaudio on Hugging Face, from 4.6B to 35.9B parameters, published by fishaudio in BF16 and F8_E4M3, with a 256K-token context. The smallest official model, s2-pro, needs about 10.2 GB of VRAM at its published precision; the cheapest live fit is RTX 3060 at $0.110/hr.
Pick a size
One row per official fishaudio 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 |
|---|---|---|---|---|---|---|---|
| s2-pro | 4.6B | 10.2 GB | 5.1 GB | 2.5 GB | RTX 3060 | $0.110/hr | 18.0 GB |
| drama-3 | 35.9B | 40.1 GB | – | 20.1 GB | L40 | $0.759/hr | not published for this architecture |
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 (2)
| Model | Parameters | Published as | VRAM needed | Live GPU fit | Est. $/hr |
|---|---|---|---|---|---|
| s2-pro | 4.6B | BF16 | 10.2 GB | RTX 3060 | $0.110/hr |
| drama-3 | 35.9B | F8_E4M3 | 40.1 GB | L40 | $0.759/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.