Audio8 models
2 Audio8 models from Audio8 on Hugging Face, from 170M to 4.1B parameters, published by Audio8 in BF16, with contexts from 8K to 32K tokens. The smallest official model, Audio8-TTS-Preview-0.1b, needs about 0.4 GB of VRAM at its published precision; the cheapest live fit is RTX 3070 at $0.088/hr.
Pick a size
One row per official Audio8 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 |
|---|---|---|---|---|---|---|---|
| Audio8-TTS-Preview-0.1b | 170M | 0.4 GB | 0.2 GB | 0.1 GB | RTX 3070 | $0.088/hr | not published for this architecture |
| ARK-ASR-3B | 4.1B | 9.1 GB | 4.5 GB | 2.3 GB | RTX 4070 | $0.154/hr | 1.1 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 (2)
| Model | Parameters | Published as | VRAM needed | Live GPU fit | Est. $/hr |
|---|---|---|---|---|---|
| Audio8-TTS-Preview-0.1b | 170M | BF16 | 0.4 GB | RTX 3070 | $0.088/hr |
| ARK-ASR-3B | 4.1B | BF16 | 9.1 GB | RTX 4070 | $0.154/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.