Check my GPU

on

Can the Intel Arc Pro B50 run Gemma 4 12B?

83 ptsGreat (A) · 83 of 100 points: Model quality 35.5/45, Memory fit 25.0/25, Speed 9.9/15, Memory use 13.5/15

Yes, it runs well

Runs well. At Q4_K_M it needs about 9.4 GB of 16.0 GB usable, with an estimated speed of 17.6 tok/s.

Best quant
Q4_K_M
Generation speed
17.6 tok/s
Prompt processing
639 tok/s
First token (1K prompt)
1.6 s
Largest safe context
87K

Download Gemma 4 12B

Recommended for the Intel Arc Pro B50: Q4_K_M. Every other quantization has its own download link in the table below.

Quickest · Ollama

One command downloads a ready-to-run version and starts a chat.

ollama run gemma4:12b-it-q4_K_M

Get Ollama at ollama.com/download (opens in a new tab) · model page (opens in a new tab)

Quantized files · GGUF

Works in LM Studio, llama.cpp, Jan and Ollama. No sign-in needed.

Download the Q4_K_M file (opens in a new tab)

huggingface.co/unsloth/gemma-4-12B-it-GGUF (opens in a new tab)

ollama run hf.co/unsloth/gemma-4-12B-it-GGUF:Q4_K_M

Original weights · Hugging Face

Full-precision files from Google, for fine-tuning, vLLM, Transformers.

huggingface.co/google/gemma-4-12B-it (opens in a new tab)

No sign-in needed.

Official repo · GitHub

Reference code, inference scripts and docs from Google. The weights themselves download from Hugging Face.

github.com/google-deepmind/gemma (opens in a new tab)

git clone https://github.com/google-deepmind/gemma

Google DeepMind’s official Gemma library.

Links checked 5 days ago against Hugging Face, Ollama and GitHub.

How the A grade is scored

Every result gets up to 100 points from four parts. The points decide the letter grade.

Model qualityEditorial quality tier of the model, scaled by the quantization or precision used.35.5 / 45
Memory fitFull marks when everything sits in GPU or unified memory; fewer when part spills to system RAM.25.0 / 25
SpeedGeneration speed against a 40 tokens/s target for LLMs, or render time against a target for image and video.9.9 / 15
Memory useRewards using the memory you have, so a bigger, better variant beats a tiny one that leaves most memory idle.13.5 / 15
Total83.9 / 10083 ptsGreat (A) · 83 of 100 points: Model quality 35.5/45, Memory fit 25.0/25, Speed 9.9/15, Memory use 13.5/15

Grade scale: S 85+ · A 72+ · B 58+ · C 42+ · D 25+ · F under 25 points.

Every quantization on the Intel Arc Pro B50

Quant Memory needed Available Speed Fit Grade · points Download
Q2_K 6.9 GB 16.0 GB 28.8 tok/s Runs well 73 ptsGreat (A) · 73 of 100 points: Model quality 26.1/45, Memory fit 25.0/25, Speed 12.7/15, Memory use 9.9/15 GGUF (UD-Q2_K_XL) (opens in a new tab)
Q3_K_M 7.9 GB 16.0 GB 22.7 tok/s Runs well 79 ptsGreat (A) · 79 of 100 points: Model quality 31.4/45, Memory fit 25.0/25, Speed 11.3/15, Memory use 11.4/15 GGUF (opens in a new tab)
Q4_K_M 9.4 GB 16.0 GB 17.6 tok/s Runs well 83 ptsGreat (A) · 83 of 100 points: Model quality 35.5/45, Memory fit 25.0/25, Speed 9.9/15, Memory use 13.5/15 GGUF (opens in a new tab)
Q5_K_M 10.6 GB 16.0 GB 14.5 tok/s Runs well 85 ptsExcellent (S) · 85 of 100 points: Model quality 36.2/45, Memory fit 25.0/25, Speed 9.0/15, Memory use 15.0/15 GGUF (opens in a new tab)
Q6_K 11.9 GB 16.0 GB 12.2 tok/s Runs well 85 ptsExcellent (S) · 85 of 100 points: Model quality 37.0/45, Memory fit 25.0/25, Speed 8.3/15, Memory use 15.0/15 GGUF (opens in a new tab)
Q8_0 14.7 GB 16.0 GB 8.9 tok/s Tight fit 83 ptsGreat (A) · 83 of 100 points: Model quality 37.4/45, Memory fit 24.0/25, Speed 7.1/15, Memory use 15.0/15 GGUF (opens in a new tab)
F16 26.0 GB 16.0 GB 2.5 tok/s Heavy offload 60 ptsGood (B) · 60 of 100 points: Model quality 37.4/45, Memory fit 4.0/25, Speed 3.7/15, Memory use 15.0/15 GGUF (opens in a new tab) Original weights (opens in a new tab)

Memory figures assume 8K of context. Longer conversations or documents grow the KV cache.

Other hardware that runs it

Estimates come from published hardware specs and our open formulas, not from running the model on your machine. Real speed varies with runtime, drivers and settings. Numbers marked “Measured” come from a linked benchmark.