Can I run Gemma 3 12B on the Apple M4 Pro?
Runs greatQ4_K_M fits in unified memory (8.4 of 16.8 GB usable by the GPU): Runs great, est. 20.9 tok/s (16.7–25.1, calibrated estimate ±20%).
24 GB unified memory · 273.0 GB/s · FP16 17.0 TFLOPS. Assumes 8k context and an f16 KV cache. Verdicts are shown for each memory size.
Verdict by memory size
The page uses the 24 GB configuration; here is every size.
- 24 GBRuns greatest. 20.9 tok/s (16.7–25.1, calibrated estimate ±20%)
- 48 GBRuns greatest. 20.9 tok/s (16.7–25.1, calibrated estimate ±20%)
- 64 GBRuns greatest. 20.9 tok/s (16.7–25.1, calibrated estimate ±20%)
Every quant of Gemma 3 12B on the Apple M4 Pro
| Quant | File size | Verdict | Speed | Memory | Runs as | Notes |
|---|---|---|---|---|---|---|
| Q2_K | 4.77 GB | Runs great | est. 30.9 tok/s24.7–37.0calibrated estimate ±20% | 5.9 / 16.8 GB | Unified memory | — |
| IQ4_XS | 6.55 GB | Runs great | est. 23.1 tok/s18.5–27.7calibrated estimate ±20% | 7.7 / 16.8 GB | Unified memory | — |
| Q4_K_Mbaseline | 7.30 GB | Runs great | est. 20.9 tok/s16.7–25.1calibrated estimate ±20% | 8.4 / 16.8 GB | Unified memory | — |
| Q8_0 | 12.51 GB | Runs well | est. 12.6 tok/s10.0–15.1calibrated estimate ±20% | 13.6 / 16.8 GB | Unified memory | — |
Where the memory goes at Q4_K_M
- Weights
- 7.3 GB
- KV cache
- 0.5 GB
- Compute buffer
- 0.6 GB
- Free
- 8.4 GB
- Not usable by the GPU
- 7.2 GB
Context length vs. KV cache
| Context | KV f16 | KV q8_0 | KV q4_0 |
|---|---|---|---|
| 4k | 8.1 GBRuns great est. 21.6 tok/s17.3–26.0calibrated estimate ±20% | 8.0 GBRuns great est. 22.0 tok/s17.6–26.4calibrated estimate ±20% | 7.9 GBRuns great est. 22.2 tok/s17.8–26.6calibrated estimate ±20% |
| 8k | 8.4 GBRuns great est. 20.9 tok/s16.7–25.1calibrated estimate ±20% | 8.2 GBRuns great est. 21.6 tok/s17.3–25.9calibrated estimate ±20% | 8.0 GBRuns great est. 22.0 tok/s17.6–26.4calibrated estimate ±20% |
| 16k | 9.0 GBRuns well est. 19.6 tok/s15.6–23.5calibrated estimate ±20% | 8.5 GBRuns great est. 20.8 tok/s16.6–25.0calibrated estimate ±20% | 8.3 GBRuns great est. 21.5 tok/s17.2–25.8calibrated estimate ±20% |
| 32k | 10.2 GBRuns well est. 17.3 tok/s13.9–20.8calibrated estimate ±20% | 9.2 GBRuns well est. 19.4 tok/s15.5–23.3calibrated estimate ±20% | 8.7 GBRuns great est. 20.7 tok/s16.6–24.8calibrated estimate ±20% |
| 64k | 12.7 GBRuns well est. 14.1 tok/s11.3–17.0calibrated estimate ±20% | 10.7 GBRuns well est. 17.1 tok/s13.7–20.5calibrated estimate ±20% | 9.6 GBRuns well est. 19.2 tok/s15.4–23.1calibrated estimate ±20% |
| 128k | 15.9 GBRuns slowly est. 6.0 tok/s4.2–7.8theoretical estimate ±30% | 13.6 GBRuns well est. 13.8 tok/s11.0–16.5calibrated estimate ±20% | 11.4 GBRuns well est. 16.8 tok/s13.4–20.2calibrated estimate ±20% |
S = Runs great · A = Runs well · B = Runs slowly · F = Won't run
At Q4_K_M the verdict stays Runs great up to 8k context with an f16 KV cache, and up to 16k with q8_0.
Estimated speed
- Generation
- est. 20.9 tok/s (16.7–25.1, calibrated estimate ±20%)
- Prompt processing
- about 765 tok/s (459–1,071, rough estimate ±40%)
Measured on this exact combination
No public measurement for this exact combination yet; the numbers above are estimates.
If this is not enough
A larger model that also runs well on the Apple M4 Pro
Runs wellMistral Small 3.2 24B: Runs well, est. 10.5 tok/s (8.4–12.5, calibrated estimate ±20%)
How to run it
Commands for the Q4_K_M file. Both tools download from Hugging Face on first run.
File: google_gemma-3-12b-it-Q4_K_M.gguf (7.30 GB) from bartowski/google_gemma-3-12b-it-GGUF.
llama-server -hf bartowski/google_gemma-3-12b-it-GGUF:Q4_K_M -c 8192 -ngl all -fa on- -ngl all loads every layer on the GPU.
- -fa on enables flash attention (needed for KV cache quantization).
- --cache-type-k q8_0 --cache-type-v q8_0 shrinks the KV cache to 0.29 GB at 8k context.
Terminal:
OLLAMA_CONTEXT_LENGTH=8192 ollama serve
ollama run hf.co/bartowski/google_gemma-3-12b-it-GGUF:Q4_K_M- Ollama defaults to 4k context on GPUs under 24 GB. Set OLLAMA_CONTEXT_LENGTH when starting the server, or /set parameter num_ctx inside the chat.
/set parameter num_ctx 8192
Related pages
Guide: Running local LLMs on a Mac (M4, M5): how much unified memory?
Other models on the Apple M4 Pro
Gemma 3 12B on other hardware
Frequently asked questions
Can I run Gemma 3 12B on the Apple M4 Pro?
Q4_K_M fits in unified memory (8.4 of 16.8 GB usable by the GPU): Runs great, est. 20.9 tok/s (16.7–25.1, calibrated estimate ±20%). Q4_K_M needs 8.4 GB at 8k context; the GPU can use 16.8 GB of the unified memory.
How much context can Gemma 3 12B use on the Apple M4 Pro?
At Q4_K_M the verdict stays Runs great up to 8k context with an f16 KV cache (0.5 GB of KV) and up to 16k with a q8_0 KV cache (0.6 GB). The model supports up to 128k.
Which quant should I use, and how fast is it?
Q4_K_M (7.30 GB) is the recommended balance: Runs great, est. 20.9 tok/s (16.7–25.1, calibrated estimate ±20%). It is also the largest tracked file that gets this verdict.
Speeds are estimates from memory bandwidth, calibrated against public benchmarks, and each one comes with an error band and a confidence label. Real results vary with drivers, backend, context length and thermals.