Aya Expanse 32B — Hardware Requirements & GPU Compatibility
Chat
Aya Expanse 32B is Cohere Labs' 32.3-billion-parameter multilingual chat model, built on the Command-R model family and refined through data arbitrage, multilingual preference training, safety tuning, and model merging aimed at closing the performance gap between English and lower-resource languages. It is optimized to perform well across 23 languages including Arabic, Chinese, French, Hindi, Japanese, and Russian, and Cohere reported it outperforming much larger models such as Llama 3.1 405B and Mistral Large 2 on multilingual evaluation. At 32.3 billion parameters, it needs a high-end consumer GPU or multi-GPU setup once quantized. It is released under the CC BY-NC 4.0 license, restricting use to non-commercial purposes only, and was published in October 2024, as the 32B counterpart to the smaller Aya Expanse 8B model.
5.7K downloads
302 likesSpecifications
- Publisher
- Cohere
- Family
- Aya
- Parameters
- 32.3B
- Release Date
- 2024-10-23
- License
- CC BY-NC 4.0
Get Started
HuggingFace
How Much VRAM Does Aya Expanse 32B Need?
Select a quantization to see compatible GPUs below.
| Quantization | Bits | VRAM | + Context | File Size | Quality |
|---|---|---|---|---|---|
| Q2_Kest. | 3.40 | 15.1 GB | — | 13.73 GB | 2-bit quantization with K-quant improvements |
| Q3_K_Mest. | 3.90 | 17.3 GB | — | 15.74 GB | 3-bit medium quantization |
| Q4_K_Mest. | 4.80 | 21.3 GB | — | 19.38 GB | 4-bit medium quantization — most popular sweet spot |
| Q5_K_Mest. | 5.70 | 25.3 GB | — | 23.01 GB | 5-bit medium quantization — good quality/size tradeoff |
| Q6_Kest. | 6.60 | 29.3 GB | — | 26.64 GB | 6-bit quantization, very good quality |
| Q8_0est. | 8.00 | 35.5 GB | — | 32.30 GB | 8-bit quantization, near-lossless |
| BF16est. | 16.00 | 71.0 GB | — | 64.59 GB | Brain floating point 16 — preferred for training |
est.= calculated VRAM estimate; no published GGUF file found for that quantization yet. Other rows are verified against real community uploads.
Which GPUs Can Run Aya Expanse 32B?
Q4_K_M · 21.3 GBAya Expanse 32B (Q4_K_M) requires 21.3 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 28+ GB is recommended. 7 GPUs can run it, including NVIDIA GeForce RTX 5090, NVIDIA GeForce RTX 3090 Ti.
Runs great
— Plenty of headroom
Which Devices Can Run Aya Expanse 32B?
Q4_K_M · 21.3 GB41 devices with unified memory can run Aya Expanse 32B, including NVIDIA DGX H100, NVIDIA DGX A100 640GB, Mac Mini M4 Pro (24 GB).
Runs great
— Plenty of headroomDecent
— Enough memory, may be tightBenchmarks
Benchmark details →Related Models
Frequently Asked Questions
- How much VRAM does Aya Expanse 32B need?
Aya Expanse 32B requires 21.3 GB of VRAM at Q4_K_M, or 71.0 GB at BF16.
VRAM = Weights + KV Cache + Overhead
Weights = 32.3B × 4.8 bits ÷ 8 = 19.4 GB
KV Cache + Overhead ≈ 1.9 GB (at 2K context + ~0.3 GB framework)
VRAM usage by quantization
Q4_K_M21.3 GB- Can NVIDIA GeForce RTX 4090 run Aya Expanse 32B?
Yes, at Q4_K_M (21.3 GB) or lower. Higher quantizations like Q5_K_M (25.3 GB) exceed the NVIDIA GeForce RTX 4090's 24 GB.
- What's the best quantization for Aya Expanse 32B?
For Aya Expanse 32B, Q4_K_M (21.3 GB) offers the best balance of quality and VRAM usage. Q5_K_M (25.3 GB) provides better quality if you have the VRAM. The smallest option is Q2_K at 15.1 GB.
VRAM requirement by quantization
Q2_K15.1 GBQ4_K_M ★21.3 GBQ5_K_M25.3 GBQ6_K29.3 GBQ8_035.5 GBBF1671.0 GB★ Recommended — best balance of quality and VRAM usage.
- Can I run Aya Expanse 32B on a Mac?
Aya Expanse 32B requires at least 15.1 GB at Q2_K, which exceeds the unified memory of most consumer Macs. You would need a Mac Studio or Mac Pro with a high-memory configuration.
- Can I run Aya Expanse 32B locally?
Yes — Aya Expanse 32B can run locally on consumer hardware. At Q4_K_M quantization it needs 21.3 GB of VRAM. Popular tools include Ollama, LM Studio, and llama.cpp.
- How fast is Aya Expanse 32B?
At Q4_K_M, Aya Expanse 32B can reach ~225 tok/s on AMD Instinct MI350X. On NVIDIA GeForce RTX 4090: ~31 tok/s. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.
tok/s = (bandwidth GB/s ÷ model GB) × efficiency
Example: NVIDIA B200 → 8000 ÷ 21.3 × 0.65 = ~244 tok/s
Estimated speed at Q4_K_M (21.3 GB)
~244 tok/s~31 tok/s~244 tok/s~225 tok/sReal-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
- What's the download size of Aya Expanse 32B?
At Q4_K_M, the download is about 19.38 GB. The full-precision BF16 version is 64.59 GB. The smallest option (Q2_K) is 13.73 GB.
- Which GPUs can run Aya Expanse 32B?
7 consumer GPUs can run Aya Expanse 32B at Q4_K_M (21.3 GB). Top options include NVIDIA GeForce RTX 5090, AMD Radeon RX 7900 XTX, NVIDIA GeForce RTX 3090. 1 GPU have plenty of headroom for comfortable inference.
- Which devices can run Aya Expanse 32B?
41 devices with unified memory can run Aya Expanse 32B at Q4_K_M (21.3 GB), including AMD Ryzen AI 9 HX 370 (Strix Point) Laptop, ASUS Ascent GX10, Asus ROG Flow Z13 (2025, Ryzen AI Max+ 395, 128 GB), Beelink GTR9 Pro (Ryzen AI Max+ 395, 128 GB). Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.