Can GPT-OSS 20B run on Intel Arc A730M 12GB?

YES — With Q2_K

A79Great

Estimated from fit model

GPT-OSS 20B needs ~12.7 GB VRAM. Intel Arc A730M 12GB has 12.0 GB. With Q2_K quantization, expect ~28 tok/s.

Runtime: llama.cppCapacity: OffloadBandwidth: LowStack: StandardBottleneck: Host offload

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.

GPT-OSS 20B at Q4_K_M needs 17.4 GB — too much for Intel Arc A730M 12GB (12.0 GB). Runs at Q2_K (12.7 GB) with low quality.

Capabilities:

Select quantization to explore

Q4_K_M (Medium quality) — 17.4 GB, exceeds 12.0 GB available

17.4 GB required12.0 GB available

145% VRAM needed

5.4 GB over capacity — needs offload or smaller quantization

Fit status

Too heavy

Decode

10.9 tok/s

TTFT

17750 ms

Safe context

Memory

17.4 GB / 12.0 GB

Offload

30%

Memory breakdown

Weights12.8 GB

KV Cache2.4 GB

Runtime0.9 GB

Headroom1.2 GB

See how fast it feels

With memory offload — actual speed may be lower

See how fast it feelsGPT-OSS 20B on Intel Arc A730M 12GB
1st promptCold start — includes initialization
>Write a Python function to sort a list
Here's a Python function that sorts a list using quicksort:

def quicksort(arr):
    if len(arr) <= 1:
        return arr
    pivot = arr[len(arr) // 2]
    left   = [x for x in arr if x < pivot]
    middle = [x for x in arr if x == pivot]
    right  = [x for x in arr if x > pivot]
    return quicksort(left) + middle + quicksort(right)

# Example usage
numbers = [3, 6, 8, 10, 1, 2, 1]
print(quicksort(numbers))  # [1, 1, 2, 3, 6, 8, 10]

The algorithm picks a pivot, partitions the list into smaller, equal, and larger elements, then recursively sorts each partition and concatenates the results.
2nd promptWarm — model ready, no init overhead
>Now add type hints and a docstring
def quicksort(arr: list[int]) -> list[int]:
    """Sort a list using the quicksort algorithm.

    Args:
        arr: List of integers to sort.

    Returns:
        A new sorted list.
    """
    if len(arr) <= 1:
        return arr
    pivot = arr[len(arr) // 2]
    left   = [x for x in arr if x < pivot]
    middle = [x for x in arr if x == pivot]
    right  = [x for x in arr if x > pivot]
    return quicksort(left) + middle + quicksort(right)

Added type hints for the input and return type, plus a Google-style docstring.
Estimated: 10.9 tok/s decode · 17.8s TTFT (warm) · 27 tok/s prefill

What limits this setup

It fits through host-memory offload, and offload is the main reason performance drops.

CPU or host-memory offload is active

About 10% of the working set spills out of accelerator memory, which usually hurts latency and sustained decode throughput.

Very little memory headroom

You can run the model, but there is not much room left for longer context, bigger batches, extra apps, or future model updates.

Runtime ecosystem is narrower than CUDA

Intel GPUs can look attractive on memory per dollar, but local AI tooling, kernels, and model coverage are still broader and easier on CUDA today.

Best improvement path

Remove offload with more accelerator memory

Prioritize a GPU or unified-memory tier that fits the whole model natively. Removing offload usually helps more than small compute gains.

Prefer CUDA if you want the path of least resistance

If your goal is maximum runtime coverage, easier troubleshooting, and better support for new local AI releases, CUDA is usually still the safer upgrade path.

Buy headroom, not only minimum fit

A slightly larger memory tier gives you safer context growth and makes the recommendation more future-proof.

Performance by workload

Workload	Grade	Fit	Decode	TTFT	Context
Chat	F	Too heavy	12.7 tok/s	8304 ms	4K
Coding	F	Too heavy	10.9 tok/s	17750 ms	4K
Agentic Coding	F	Too heavy	8.3 tok/s	34062 ms	4K
Reasoning	F	Too heavy	10.9 tok/s	20977 ms	4K
RAG	F	Too heavy	8.3 tok/s	42578 ms	4K

Inference speed

GPT-OSS 20B inference speed — tokens per second by GPU & Mac

Estimated decode speed (tokens/sec) for GPT-OSS 20B at Q4_K_M across popular GPUs and Apple Silicon, using the fastest local runtime per device. Fastest is RTX 5090 32GB at ~231 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 / Mac	Memory	Quant	Speed (tok/s)	Fits?
RTX 5090 32GB	32 GB	Q4_K_M	230.5	Fits
RTX 4090 24GB	24 GB	Q4_K_M	147.1	Fits
RX 7900 XTX 24GB	24 GB	Q4_K_M	132.7	Fits
RTX 3090 24GB	24 GB	Q4_K_M	125.8	Fits
Mac Studio M3 Ultra 256GB	256 GB	Q4_K_M	106.9	Fits
Mac Studio M2 Ultra 128GB	128 GB	Q4_K_M	89.1	Fits
Mac Studio M1 Ultra 128GB	128 GB	Q4_K_M	84.5	Fits
RTX 4080 Super 16GB	16 GB	Q4_K_M	68.2	Heavy offload
MacBook Pro M4 Max 128GB	128 GB	Q4_K_M	66.0	Fits
MacBook Pro M4 Max 64GB	64 GB	Q4_K_M	66.0	Fits
MacBook Pro M3 Max 64GB	64 GB	Q4_K_M	46.1	Fits
MacBook Pro M1 Max 64GB	64 GB	Q4_K_M	42.2	Fits
MacBook Pro M4 Pro 48GB	48 GB	Q4_K_M	40.4	Fits
RTX 4070 12GB	12 GB	Q4_K_M	24.2	Too big
RTX 3060 12GB	12 GB	Q4_K_M	15.2	Too big
RTX 4060 8GB	8 GB	Q4_K_M	5.7	Too 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 GPT-OSS 20B (21B params) fits at each quantization level on Intel Arc A730M 12GB (12.0 GB usable).

Quant	Bits	VRAM	Quality	Fit
Q2_KBest for your GPU	2	8.2 GB	Low	S90
Q3_K_S	3	10.3 GB	Low	F0
NVFP4	4	11.8 GB	Medium	F0
Q4_K_M	4	12.8 GB	Medium	F0
Q5_K_M	5	15.1 GB	High	F0
Q6_K	6	17.2 GB	High	F0
Q8_0	8	22.5 GB	Very High	F0
F16	16	43.1 GB	Maximum	F0

Get started

Copy-paste commands to run GPT-OSS 20B on your machine.

Run

ollama run gpt-oss

Opções de upgrade

Hardware que roda bem GPT-OSS 20B

Intel Arc A770 16GBMelhor custo-benefício

16 GB VRAM (+4)560 GB/s (+224)

Makes the model fit on the accelerator instead of staying completely out of reach.29.2 tok/s decodificação

Makes the model fit on the accelerator instead of staying completely out of reach.

Raises estimated decode speed by about 168%.

~$349 MSRP

Intel Arc Pro B50 16GBUpgrade Intel

16 GB VRAM (+4)

Makes the model fit on the accelerator instead of staying completely out of reach.14.4 tok/s decodificação

Makes the model fit on the accelerator instead of staying completely out of reach.

Raises estimated decode speed by about 32%.

~$399 MSRP

Intel Arc Pro B60 24GBOpção econômica

24 GB VRAM (+12)456 GB/s (+120)

Makes the model fit on the accelerator instead of staying completely out of reach.47.3 tok/s decodificação

Makes the model fit on the accelerator instead of staying completely out of reach.

Removes host-memory offload, which is usually the single biggest latency and throughput win.

~$599 MSRP

Frequently asked questions

Can Intel Arc A730M 12GB run GPT-OSS 20B?

Yes, Intel Arc A730M 12GB can run GPT-OSS 20B at Q2_K quantization (Runs with offload (needs ~0.5 GB host RAM)). The recommended Q4_K_M requires 17.4 GB which exceeds available memory, but at Q2_K it needs only 12.7 GB. Expected decode speed: 27.8 tok/s.

How much VRAM does GPT-OSS 20B need?

GPT-OSS 20B (21B parameters) requires approximately 17.4 GB at Q4_K_M quantization. On Intel Arc A730M 12GB, it fits at Q2_K using 12.7 GB.

What is the best quantization for GPT-OSS 20B?

The recommended quantization is Q4_K_M, but on Intel Arc A730M 12GB the best fitting quantization is Q2_K, which uses 12.7 GB.

What speed will GPT-OSS 20B run at on Intel Arc A730M 12GB?

On Intel Arc A730M 12GB, GPT-OSS 20B achieves approximately 27.8 tokens per second decode speed with a time-to-first-token of 6958ms using Q2_K quantization.

Can Intel Arc A730M 12GB run GPT-OSS 20B for coding?

For coding workloads, GPT-OSS 20B on Intel Arc A730M 12GB receives a F grade with 10.9 tok/s and 4K context.

What context window can GPT-OSS 20B use on Intel Arc A730M 12GB?

On Intel Arc A730M 12GB, GPT-OSS 20B can safely use up to 11K tokens of context at Q2_K quantization. The model's official context limit is 128K, but available memory constrains the safe maximum.

What should I upgrade first if GPT-OSS 20B feels slow on Intel Arc A730M 12GB?

Remove offload with more accelerator memory. Prioritize a GPU or unified-memory tier that fits the whole model natively. Removing offload usually helps more than small compute gains.

Would CUDA be a better path than Intel Arc A730M 12GB for GPT-OSS 20B?

Often yes, if your goal is the easiest setup and the widest runtime support. Intel can offer attractive memory capacity, but CUDA still tends to win on tooling maturity, guides, kernels, and model coverage for local AI.

See all results for Intel Arc A730M 12GB See all hardware for GPT-OSS 20B

Embed this result▼

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

<iframe src="https://willitrunai.com/embed/gpt-oss-20b-on-arc-a730m-12gb" width="400" height="180" frameborder="0" style="border:none;border-radius:12px;overflow:hidden;" title="Will It Run AI — fit result"></iframe>

Preview: