Aya Expanse 32B
by Cohere · aya family
32B
parameters
text-generation multilingual creative-writing summarization
Aya Expanse 32B is Cohere's flagship multilingual model, covering 23 languages with higher quality than its 8B sibling. It delivers strong multilingual generation, translation, and comprehension. At Q4 it needs about 22 GB VRAM, fitting on a RTX 3090/4090 or a Mac with 24 GB+ unified memory. The best open multilingual model for users who work across multiple languages.
Quick Start with Ollama
ollama run 32b-q4_K_M | Creator | Cohere |
| Parameters | 32B |
| Architecture | transformer-decoder |
| Context | 8K tokens |
| Released | Oct 15, 2025 |
| License | CC-BY-NC-4.0 |
| Ollama | aya-expanse:32b |
Quantization Options
| Format | File Size | VRAM Required | Quality | Ollama Tag |
|---|---|---|---|---|
| Q4_K_M rec | 20 GB | 22 GB | | 32b-q4_K_M |
| Q8_0 | 35 GB | 38 GB | | 32b-q8_0 |
Compatible Hardware
Q4_K_M requires 22 GB VRAM