What fits in 48 GB of unified memory
Data updated . Last reviewed the same day, when this catalog snapshot was written.
2 catalog machines have 48 GB unified. Models are ranked against the Mac mini M4 Pro 48GB, the tightest usable pool at this sticker size (33.6 GB). The tok/s estimate uses Mac mini M4 Pro 48GB bandwidth (273 GB/s) and is not a benchmark. 1 of these fits do not clear the 36 GB tier.
Models
| Model | Best quant | Estimate | Headroom | Est. tok/s |
|---|---|---|---|---|
| Llama 3.2 1B Instruct | FP16 | 3.0 GB | 30.6 GB | 110 tok/s |
| Llama 3.2 3B Instruct | FP16 | 7.7 GB | 25.9 GB | 42.5 tok/s |
| Phi-3.5 mini | FP16 | 9.2 GB | 24.4 GB | 35.7 tok/s |
| Phi-4 mini | FP16 | 9.2 GB | 24.4 GB | 35.6 tok/s |
| Gemma 3 4B | FP16 | 10.3 GB | 23.3 GB | 31.7 tok/s |
| Mistral 7B Instruct v0.3 | FP16 | 17.4 GB | 16.2 GB | 18.8 tok/s |
| Qwen2.5 7B Instruct | FP16 | 18.3 GB | 15.3 GB | 17.9 tok/s |
| Gemma 4 E4B | FP16 | 19.2 GB | 14.4 GB | 17.1 tok/s |
| DeepSeek R1 Distill Llama 8B | FP16 | 19.3 GB | 14.3 GB | 17.0 tok/s |
| Llama 3.1 8B Instruct | FP16 | 19.3 GB | 14.3 GB | 17.0 tok/s |
| Qwen3 8B | FP16 | 19.7 GB | 13.9 GB | 16.7 tok/s |
| Ministral 3 8B | FP16 | 21.4 GB | 12.2 GB | 15.3 tok/s |
| Gemma 2 9B | FP16 | 22.2 GB | 11.4 GB | 14.8 tok/s |
| Gemma 4 12B | FP16 | 28.7 GB | 4.9 GB | 11.4 tok/s |
| Gemma 3 12B | FP16 | 29.2 GB | 4.4 GB | 11.2 tok/s |
| Mistral Nemo 12B | FP16 | 29.4 GB | 4.2 GB | 11.1 tok/s |
| Ministral 3 14B | FP16 | 33.5 GB | 0.1 GB | 9.79 tok/s |
| DeepSeek V4 Flash | Q8 | 4.5 GB | 29.1 GB | 1605 tok/s |
| gpt-oss-20b | Q8 | 12.3 GB | 21.3 GB | 131 tok/s |
| Qwen3 14B | Q8 | 17.0 GB | 16.6 GB | 17.4 tok/s |
| Phi-4 | Q8 | 17.3 GB | 16.3 GB | 17.5 tok/s |
| Qwen2.5 14B Instruct | Q8 | 17.3 GB | 16.3 GB | 17.4 tok/s |
| Mistral Small 3.2 24B | Q8 | 26.4 GB | 7.2 GB | 10.9 tok/s |
| Gemma 4 26B-A4B | Q8 | 27.2 GB | 6.4 GB | 69.0 tok/s |
| Gemma 3 27B | Q8 | 28.7 GB | 4.9 GB | 9.51 tok/s |
| Qwen3.6 27B | Q8 | 29.1 GB | 4.5 GB | 9.55 tok/s |
| Qwen3.8 27B | Q8 | 29.6 GB | 4.0 GB | 9.40 tok/s |
| Gemma 2 27B | Q8 | 30.0 GB | 3.6 GB | 9.09 tok/s |
| GLM-4.7-Flash | Q8 | 32.3 GB | 1.3 GB | 89.2 tok/s |
| Qwen3 30B-A3B | Q8 | 33.3 GB | 0.3 GB | 85.5 tok/s |
| Gemma 4 31B | Q4 | 19.7 GB | 13.9 GB | 14.9 tok/s |
| Qwen3 32B | Q4 | 21.9 GB | 11.7 GB | 13.8 tok/s |
| DeepSeek R1 Distill Qwen 32B | Q4 | 22.0 GB | 11.6 GB | 13.8 tok/s |
| Qwen2.5 32B Instruct | Q4 | 22.0 GB | 11.6 GB | 13.8 tok/s |
| Qwen3.6 35B-A3B | Q4 | 22.3 GB | 11.3 GB | 148 tok/s |
| Mixtral 8x7B | Q4 | 27.5 GB | 6.1 GB | 37.4 tok/s |
Machines at 48 GB
- Mac mini M4 Pro 48GB
33.6 GB usable · $1,799
- MacBook Pro 14-inch M4 Pro 48GB
33.6 GB usable · $2,399
Questions
- How many models fit in 48 GB of unified memory?
- 36 of 49 fit on the tightest config at this size, the Mac mini M4 Pro 48GB, which keeps 33.6 GB usable.
- Which catalog machines have 48 GB unified?
- Mac mini M4 Pro 48GB (70% usable), MacBook Pro 14-inch M4 Pro 48GB (70% usable).
- Why is the ranking based on the tightest config?
- So a model listed as fitting works on every 48 GB machine here, including ones that reserve more memory for the OS. A looser machine has more headroom than the table shows.