willitrun·ai

Can Aya Expanse 32B run on AMD Instinct MI350X 288GB?

YES — Runs Great

C51Usable
Estimated from fit model

Aya Expanse 32B needs ~51.7 GB VRAM. AMD Instinct MI350X 288GB has 288.0 GB. With Q4_K_M quantization, expect ~325 tok/s.

Runtime: llama.cppCapacity: RoomyBandwidth: HighStack: StandardBottleneck: Balanced
Share:

Operating mode

Choose the run profile you care about

Interactive favors responsiveness, while light API and scale-out lean harder on serving readiness. The fit stays the same, but the recommendation lens changes.

Current mode

Balanced

Balanced for general local use. Keeps the ranking neutral across personal and serving workflows.

Capabilities:

Select quantization to explore

Q4_K_M (Medium quality) — 51.7 GB, 325.4 tok/s, Runs well
51.7 GB required288.0 GB available
18% VRAM used

Fit status

Runs well

Decode

325.4 tok/s

TTFT

595 ms

Safe context

8K

Memory

51.7 GB / 288.0 GB

Memory breakdown

Weights19.5 GB
KV Cache2.4 GB
Runtime0.9 GB
Headroom28.8 GB

See how fast it feels

See how fast it feelsAya Expanse 32B on AMD Instinct MI350X 288GB
1st promptCold start — includes initialization
>What is local AI inference?
Local AI inference means running an AI language model directly on your own hardware — your laptop, desktop, or server — instead of sending requests to a remote cloud API. When you run inference locally the model weights are loaded into your GPU or unified memory. Each token you generate requires reading those weights from memory, so memory bandwidth is the main bottleneck for decode speed. Key benefits of running locally: - Full privacy: your prompts never leave your machine - No per-token cost or rate limits - Works offline once the model is downloaded - Latency depends only on your hardware
2nd promptWarm — model ready, no init overhead
>How much VRAM do I need?
It depends on the model size and quantization level. A rough rule of thumb: Model size Q4 (4-bit) Q8 (8-bit) FP16 7B params ~4.3 GB ~7.5 GB ~14 GB 13B params ~7.9 GB ~13.9 GB ~26 GB 70B params ~42.7 GB ~74.9 GB ~140 GB Most people use 4-bit quantization (Q4_K_M) which gives 90-95% of full quality at a fraction of the memory. A 24 GB GPU can comfortably run most 7B-13B models.
Estimated: 325.4 tok/s decode · 595ms TTFT (warm) · 813 tok/s prefill

What limits this setup

This setup is broadly balanced for this model.

No major red flags

This recommendation has enough memory headroom and acceptable estimated speed for the selected workload.

Best improvement path

Performance by workload

WorkloadGradeFitDecodeTTFTContext
ChatCRuns well325.4 tok/s350 ms8K
CodingCRuns well325.4 tok/s595 ms8K
Agentic CodingCRuns well325.4 tok/s866 ms8K
ReasoningCRuns well325.4 tok/s703 ms8K
RAGCRuns well325.4 tok/s1082 ms8K

Inference speed

Aya Expanse 32B inference speed — tokens per second by GPU & Mac

Estimated decode speed (tokens/sec) for Aya Expanse 32B at Q4_K_M across popular GPUs and Apple Silicon, using the fastest local runtime per device. Fastest is RTX 5090 32GB at ~43 tok/s. Speed is memory-bandwidth bound, so cards that fit the whole model in VRAM run far faster than ones that offload to system RAM.

GPU / MacMemoryQuantSpeed (tok/s)Fits?
NVIDIARTX 5090 32GB
32 GBQ4_K_M43.3Fits
MacBook Pro M4 Max 128GB
128 GBQ4_K_M33.5Fits
MacBook Pro M4 Max 64GB
64 GBQ4_K_M33.5Fits
Mac Studio M3 Ultra 256GB
256 GBQ4_K_M31.0Fits
RX 7900 XTX 24GB
24 GBQ4_K_M25.9Offloads
Mac Studio M2 Ultra 128GB
128 GBQ4_K_M25.9Fits
Mac Studio M1 Ultra 128GB
128 GBQ4_K_M24.5Fits
MacBook Pro M4 Pro 48GB
48 GBQ4_K_M21.1Fits
NVIDIARTX 4090 24GB
24 GBQ4_K_M16.6Offloads
NVIDIARTX 3090 24GB
24 GBQ4_K_M15.2Offloads
MacBook Pro M3 Max 64GB
64 GBQ4_K_M13.4Fits
MacBook Pro M1 Max 64GB
64 GBQ4_K_M12.3Fits
NVIDIARTX 4080 Super 16GB
16 GBQ4_K_M10.2Too big
NVIDIARTX 4070 12GB
12 GBQ4_K_M3.6Too big
NVIDIARTX 3060 12GB
12 GBQ4_K_M2.2Too big
NVIDIARTX 4060 8GB
8 GBQ4_K_M2.0Too big

Estimates for single-stream decoding at Q4_K_M; real tokens/sec varies with prompt length, context, batch size, and runtime build. Prompt processing (prefill) is faster than the decode figures shown here.

Quantization options

How Aya Expanse 32B (32B params) fits at each quantization level on AMD Instinct MI350X 288GB (288.0 GB usable).

QuantBitsVRAMQualityFit
Q2_K
2
12.5 GB
LowC41
Q3_K_S
3
15.7 GB
LowC42
NVFP4
4
17.9 GB
MediumC42
Q4_K_M
4
19.5 GB
MediumC42
Q5_K_M
5
23.0 GB
HighC42
Q6_K
6
26.2 GB
HighC42
Q8_0
8
34.2 GB
Very HighC43
F16Best for your GPU
16
65.6 GB
MaximumC45

Get started

Copy-paste commands to run Aya Expanse 32B on your machine.

Run

ollama run aya-expanse:32b

Frequently asked questions

Can AMD Instinct MI350X 288GB run Aya Expanse 32B?

Yes, AMD Instinct MI350X 288GB can run Aya Expanse 32B with a C grade (Runs well). Expected decode speed: 325.4 tok/s.

How much VRAM does Aya Expanse 32B need?

Aya Expanse 32B (32B parameters) requires approximately 51.7 GB of memory with Q4_K_M quantization.

What is the best quantization for Aya Expanse 32B?

The recommended quantization for Aya Expanse 32B is Q4_K_M, which balances quality and memory efficiency.

What speed will Aya Expanse 32B run at on AMD Instinct MI350X 288GB?

On AMD Instinct MI350X 288GB, Aya Expanse 32B achieves approximately 325.4 tokens per second decode speed with a time-to-first-token of 595ms using Q4_K_M quantization.

Can AMD Instinct MI350X 288GB run Aya Expanse 32B for coding?

For coding workloads, Aya Expanse 32B on AMD Instinct MI350X 288GB receives a C grade with 325.4 tok/s and 8K context.

What context window can Aya Expanse 32B use on AMD Instinct MI350X 288GB?

On AMD Instinct MI350X 288GB, Aya Expanse 32B can safely use up to 8K tokens of context. The model's official context limit is 8K, but available memory constrains the safe maximum.

See all results for AMD Instinct MI350X 288GBSee all hardware for Aya Expanse 32B
Embed this result▼

Paste this snippet into any page to show a live fit card.

<iframe src="https://willitrunai.com/embed/aya-expanse-32b-on-instinct-mi350x-288gb" width="400" height="180" frameborder="0" style="border:none;border-radius:12px;overflow:hidden;" title="Will It Run AI — fit result"></iframe>

Preview: