Would CUDA be a better path than Intel Arc A370M 4GB for Agents-A1 4B?

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.

Can Agents-A1 4B run on Intel Arc A370M 4GB?

Q: What should I upgrade first if Agents-A1 4B feels slow on Intel Arc A370M 4GB?

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.

YES — With Q4_K_M

B63Good

Estimated from fit model

Agents-A1 4B needs ~4.5 GB VRAM. Intel Arc A370M 4GB has 4.0 GB. With Q4_K_M quantization, expect ~9 tok/s.

Runtime: llama.cppCapacity: OffloadBandwidth: Very 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.

Agents-A1 4B at Q4_K_M needs 6.0 GB — too much for Intel Arc A370M 4GB (4.0 GB). Runs at Q4_K_M (4.5 GB) with medium quality. 6 quantization levels fit.

Capabilities:

Select quantization to explore

Q4_K_M (Medium quality) — 4.5 GB, 9.4 tok/s, Very compromised (needs ~0.3 GB host RAM)

4.5 GB required4.0 GB available

113% VRAM needed

0.5 GB over capacity — needs offload or smaller quantization

Fit status

Very compromised (needs ~0.3 GB host RAM)

Decode

9.4 tok/s

TTFT

20625 ms

Safe context

Memory

4.5 GB / 4.0 GB

Offload

10%

Memory breakdown

Weights2.7 GB

KV Cache0.5 GB

Runtime0.9 GB

Headroom0.4 GB

See how fast it feels

See how fast it feelsAgents-A1 4B on Intel Arc A370M 4GB
1st promptCold start — includes initialization
>Solve: If a train travels at 60 mph for 2.5 hours, how far does it go?
Let me work through this step by step.

Given information:
  - Speed = 60 mph (miles per hour)
  - Time  = 2.5 hours

Formula: Distance = Speed x Time

Calculation:
  Distance = 60 mph x 2.5 h
           = 150 miles

The train travels 150 miles.

Double-check: at 60 mph the train covers 60 miles each hour. In half an hour it covers 30 miles. So in 2.5 hours: 60 + 60 + 30 = 150 miles. Confirmed.
2nd promptWarm — model ready, no init overhead
>What if the train stops for 20 minutes halfway?
The stop doesn't change the total distance covered while moving, but it affects total trip time.

Breakdown:
  - First half: 1.25 hours at 60 mph = 75 miles
  - Stop: 20 minutes (no distance)
  - Second half: 1.25 hours at 60 mph = 75 miles

Total distance: 150 miles (same)
Total time: 2.5 h + 0.33 h = 2.83 hours
Average speed: 150 / 2.83 = 53 mph

The distance stays the same but average speed drops to 53 mph because of the stop.
Estimated: 9.4 tok/s decode · 20.6s TTFT (warm) · 24 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	7.0 tok/s	14999 ms	4K
Coding	F	Too heavy	4.8 tok/s	39973 ms	4K
Agentic Coding	F	Too heavy	2.7 tok/s	105243 ms	4K
Reasoning	F	Too heavy	4.8 tok/s	47241 ms	4K
RAG	F	Too heavy	2.7 tok/s	131553 ms	4K

Inference speed

Agents-A1 4B inference speed — tokens per second by GPU & Mac

Estimated decode speed (tokens/sec) for Agents-A1 4B at Q4_K_M across popular GPUs and Apple Silicon, using the fastest local runtime per device. Fastest is RTX 5090 32GB at ~86 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	85.5	Fits
RTX 4090 24GB	24 GB	Q4_K_M	72.0	Fits
RTX 4080 Super 16GB	16 GB	Q4_K_M	72.0

Quantization options

How Agents-A1 4B (4.5B params) fits at each quantization level on Intel Arc A370M 4GB (4.0 GB usable).

Quant	Bits	VRAM	Quality	Fit
Q1_0_G128	1.125	0.6 GB	Very Low	A82
Q2_0_G128	1.71	1.2 GB	Low	A81
Q2_KBest for your GPU

Get started

Copy-paste commands to run Agents-A1 4B on your machine.

Run

docker run --rm -it ghcr.io/ggerganov/llama.cpp:full \
  --hf-repo "InternScience/Agents-A1-4B" \
  --hf-file "Agents-A1-4B-Q4_K_M.gguf" \
  -c 4096 -ngl 99

Upgrade options

Hardware that runs Agents-A1 4B well

Intel Arc A380 6GBBudget pick

6 GB VRAM (+2)186 GB/s (+74)

Removes host-memory offload, which is usually the single biggest latency and throughput win.27 tok/s decode

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

Raises estimated decode speed by about 187%.

~$139 MSRP

Intel Arc A580 8GBBest value

8 GB VRAM (+4)512 GB/s (+400)

Removes host-memory offload, which is usually the single biggest latency and throughput win.63 tok/s decode

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

Raises estimated decode speed by about 570%.

~$179 MSRP

Intel Arc B570 10GBIntel upgrade

10 GB VRAM (+6)380 GB/s (+268)

Removes host-memory offload, which is usually the single biggest latency and throughput win.60.9 tok/s decode

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

Raises estimated decode speed by about 548%.

~$219 MSRP

Frequently asked questions

Can Intel Arc A370M 4GB run Agents-A1 4B?

Yes, Intel Arc A370M 4GB can run Agents-A1 4B at Q4_K_M quantization (Very compromised (needs ~0.3 GB host RAM)). The recommended Q4_K_M requires 6.0 GB which exceeds available memory, but at Q4_K_M it needs only 4.5 GB. Expected decode speed: 9.4 tok/s.

How much VRAM does Agents-A1 4B need?

Agents-A1 4B (4.5B parameters) requires approximately 6.0 GB at Q4_K_M quantization. On Intel Arc A370M 4GB, it fits at Q4_K_M using 4.5 GB.

What is the best quantization for Agents-A1 4B?

The recommended quantization is Q4_K_M, but on Intel Arc A370M 4GB the best fitting quantization is Q4_K_M, which uses 4.5 GB.

What speed will Agents-A1 4B run at on Intel Arc A370M 4GB?

On Intel Arc A370M 4GB, Agents-A1 4B achieves approximately 9.4 tokens per second decode speed with a time-to-first-token of 20625ms using Q4_K_M quantization.

Can Intel Arc A370M 4GB run Agents-A1 4B for coding?

For coding workloads, Agents-A1 4B on Intel Arc A370M 4GB receives a F grade with 4.8 tok/s and 4K context.

What context window can Agents-A1 4B use on Intel Arc A370M 4GB?

On Intel Arc A370M 4GB, Agents-A1 4B can safely use up to 4K tokens of context at Q4_K_M quantization. The model's official context limit is 262K, but available memory constrains the safe maximum.

See all results for Intel Arc A370M 4GB See all hardware for Agents-A1 4B

Embed this result▼

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

<iframe src="https://willitrunai.com/embed/agents-a1-4b-on-arc-a370m-4gb" width="400" height="180" frameborder="0" style="border:none;border-radius:12px;overflow:hidden;" title="Will It Run AI — fit result"></iframe>

Preview: