Rent a GPU

Can the Apple M2 Ultra (192GB) run Gemma 3 4B?

Apple M2 Ultra (192GB) × Gemma 3 4B8K context
YesRuns fully on the GPU
Memory at FP1610.3 GB of 144 GB
Weights 8.6 GBKV cache 1.1 GBRuntime 0.6 GB
QuantFP16Full quality
Memory10.3 GBof 144 GB usable
Speed~56tokens per second
  1. FP16
  2. Q8_0
  3. Q6_K
  4. Q5_K_M
  5. Q4_K_M
  6. Q3_K_M
  7. Q2_K

Yes. The Apple M2 Ultra (192GB) can run Gemma 3 4B fully in unified memory at FP16 (full quality), using about 10.3 GB of the 144 GB available, at roughly 56 tokens/second.

Try other context lengths

Run it with Ollama

With Ollama installed, one command downloads and starts Gemma 3 4B:

ollama run gemma3:4b

Ollama's default download is Q4_K_M (about 4.3 GB here). You have room for FP16, which is full quality: pick that GGUF in LM Studio or from Hugging Face.

Run it in the cloud

No Apple M2 Ultra (192GB)? Run Gemma 3 4B in the cloud.

Rent a cloud GPU by the hour. No upfront cost, stop any time.

Rent a GPU on RunPodRent a GPU on Vast.ai

Memory needed by quantization

At 8,192 tokens of context. The black line marks the 144 GB the Apple M2 Ultra (192GB) can use. macOS lets the GPU use about 144 GB of the 192 GB unified memory by default.

QuantQualityMemoryFits?Speed
FP16Full quality10.3 GBFits~56 tok/s
Q8_0Near-lossless6.3 GBFits~105 tok/s
Q6_KExcellent5.3 GBFits~136 tok/s
Q5_K_MVery good4.8 GBFits~157 tok/s
Q4_K_MGood, most popular4.3 GBFits~184 tok/s
Q3_K_MNoticeably degraded3.8 GBFits~228 tok/s
Q2_KHeavily degraded3.4 GBFits~298 tok/s

GPUs that run Gemma 3 4B comfortably

Models that run well on the Apple M2 Ultra (192GB)

Frequently asked questions

Can Apple M2 Ultra (192GB) run Gemma 3 4B?

Yes. The Apple M2 Ultra (192GB) can run Gemma 3 4B fully in unified memory at FP16 (full quality), using about 10.3 GB of the 144 GB available, at roughly 56 tokens/second.

How much VRAM does Gemma 3 4B need?

About 4.3 GB at Q4_K_M with 8,192 tokens of context, and about 10.3 GB at full FP16 precision.

How fast is Gemma 3 4B on the Apple M2 Ultra (192GB)?

Roughly 56 tokens per second at FP16, estimated from the Apple M2 Ultra (192GB)'s 800 GB/s memory bandwidth.