Aya Models — Hardware Requirements
5 Aya models from Cohere and the community, from the smallest that runs in 4.2 GB of VRAM at Q4 up to 32.3B parameters. Every row links to full quantization tables and GPU compatibility.
All Aya Models by Size
| Model | Params | Runs from (Q4) | Context | Publisher | Quant downloads |
|---|---|---|---|---|---|
| Tiny Aya Global | 3.3B | 4.2 GB | — | ||
| Tiny Aya Base | 3.3B | 4.2 GB | — | ||
| Aya 23 8B | 8.0B | 7.9 GB | — | ||
| Aya Expanse 8B | 8.0B | 7.9 GB | — | ||
| Aya Expanse 32B | 32.3B | 25.1 GB | — |
Community variants and fine-tunes(1)
| Model | Params | Runs from (Q4) | Context | Publisher | Quant downloads |
|---|---|---|---|---|---|
| Tiny Aya Water | 3.3B | 4.2 GB | — |
Frequently Asked Questions
- How much VRAM do I need to run a Aya model?
- The smallest Aya model, Tiny Aya Global, runs from 4.2 GB of VRAM at Q4_K_M. Larger family members need proportionally more — see the table above for every model.
- Which Aya models can I run on a 16 GB GPU?
- 5 of 6 Aya models fit in 16 GB of VRAM at Q4_K_M, including Aya 23 8B, Tiny Aya Global, Aya Expanse 8B.
- What is the most popular Aya model to run locally?
- Aya Expanse 32B is the most downloaded Aya model in local-friendly quantized formats. It runs from 25.1 GB of VRAM at Q4_K_M.