Apple · M5 · Laptop
Best AI Models for MacBook Pro 14" M5 (24 GB)
Runs 9B models well and fits 35B MoE; 14B and 27–31B fit but are slow (11 and 6 tok/s), and 70B doesn't fit in 20.5 GB.
| # | Model | Speed | Score |
|---|---|---|---|
| 1 | Q4_K_M · 16.5 GB | 31.7 tok/s | |
| 2 | Qwen3.5 9BAlibaba Q4_K_M · 6.6 GB | 16.9 tok/s | |
| 3 | GPT OSS 20BOpenAI Q4_K_M · 13.1 GB | 34.6 tok/s | |
| 4 | Qwen3 8BAlibaba Q4_K_M · 7.6 GB | 19.5 tok/s | |
| 5 | Ling 3.0 TinyInclusion AI Q4_K_M · 5.5 GB | 74.1 tok/s |
| # | Model | Speed | Score |
|---|---|---|---|
| 1 | Q4_K_M · 16.5 GB | 31.7 tok/s | |
| 2 | GPT OSS 20BOpenAI Q4_K_M · 13.1 GB | 34.6 tok/s | |
| 3 | Q4_K_M · 7.9 GB | 20.3 tok/s | |
| 4 | Qwen3.5 9BAlibaba Q4_K_M · 6.6 GB | 16.9 tok/s | overall |
| 5 | Qwen3.5 4BAlibaba Q4_K_M · 3.4 GB | 33.0 tok/s |
| # | Model | Speed | Score |
|---|---|---|---|
| 1 | Qwen3.5 9BAlibabaHow to run → Q4_K_M · 6.6 GB | 16.9 tok/s | |
| 2 | GPT OSS 20BOpenAI Q4_K_M · 13.1 GB | 34.6 tok/s | |
| 3 | Q4_K_M · 7.9 GB | 20.3 tok/s | |
| 4 | Q4_K_M · 2.3 GB | 131 tok/s | |
| 5 | Qwen3.5 4BAlibaba Q4_K_M · 3.4 GB | 33.0 tok/s |
| # | Model | Speed ↓ | Score |
|---|---|---|---|
| 1 | GPT OSS 20BOpenAIHow to run → Q4_K_M · 13.1 GB | 34.6 tok/s | |
| 2 | Ling 3.0 TinyInclusion AI Q4_K_M · 5.5 GB | 74.1 tok/s | |
| 3 | Gemma 4 26B A4B ITGoogle Q4_K_M · 16.5 GB | 31.7 tok/s | |
| 4 | Qwen3 8BAlibaba Q4_K_M · 7.6 GB | 19.5 tok/s | |
| 5 | Qwen3.5 9BAlibaba Q4_K_M · 6.6 GB | 16.9 tok/s |
| # | Model | Speed | Score |
|---|---|---|---|
| 1 | Q4_K_M · 16.5 GB | 31.7 tok/s | |
| 2 | Qwen3.5 9BAlibaba Q4_K_M · 6.6 GB | 16.9 tok/s | |
| 3 | Gemma 3 4B ITGoogle Q4_K_M · 5.0 GB | 37.9 tok/s | |
| 4 | Qwen3.5 4BAlibaba Q4_K_M · 3.4 GB | 33.0 tok/s | |
| 5 | Q4_K_M · 6.5 GB | 17.3 tok/s |
The MacBook Pro 14" M5 is a mid-range 24 GB laptop: it runs 9B models well and fits 35B MoE. Among 24 GB machines, only the M5 Pro and M4 Pro write faster. macOS keeps about 3.5 GB, leaving 20.5 GB for models. Prompt reading is its weak spot: a 4K-token prompt waits 11 s before the first word on Gemma 4 26B A4B IT. M5 generation, released 2025.
MacBook Pro 14" M5 strengths and weaknesses
Strengths
- Runs 35B MoE models: Qwen3.6 35B A3B at 31.1 tok/s in 19.8 GB
Weaknesses
- 70B models don't fit: Llama 3.3 70B Instruct needs 51.3 GB at Q4 vs 20.5 GB usable here
- Slow on large dense models: Qwen3.8 27B at 6.2 tok/s
- Slow prompt reading: A 32K-token prompt on Gemma 4 26B A4B IT waits 1.5 min for the first token
- No CUDA: vLLM, AWQ, GPTQ and FP8 builds don't run; use GGUF or MLX
Memory for models
20.5 GB of 24 GB70B models need 51.3 GB at Q4, more than the 20.5 GB available.
Other memory sizes of the same machine first, then machines and GPUs people weigh against it, each with the one reason it is worth a look. Speeds on Qwen3.5 9B.
Output speed depends on memory bandwidth (154 GB/s). Grey rows need more than the 20.5 GB available.
- 9BQwen3.5 9B · Q4_K_M · 6.6 GB16.9tok/s
- 14BQwen3 14B · Q4_K_M · 11.9 GB11.3tok/s
- 6.2tok/s
- 31.1tok/s
- 70BLlama 3.3 70B Instruct · does not fitneeds 51.3 GB
- 120B MoEGPT OSS 120B · does not fitneeds 70.8 GB
Estimated speeds: Q4_K_M, 4K context, one user. How we estimate
Chip generations: writing vs compute
M1 → M5Writing speed follows memory bandwidth and barely moved: 7.5 to 16.9 tok/s since the M1. Compute, which sets how fast prompts are read, grew 3.1×. Indexed to the M1; speeds on Qwen3.5 9B, same 16 GB.
All 1068 models on the MacBook Pro 14" M5
Featured onlyBest quant that fits in 20.5 GB, ordered by llmrun Score. Fit grades how well it runs here, S best. How ranking works
| Model | Quant | VRAM | Speed | Context | Score | Fit |
|---|---|---|---|---|---|---|
Gemma 2 9BChat | Q4_K_M | 8.8 GB | 17.6 t/s | — | B57, Decent | |
Qwen2.5 Coder 7BChatCodeFunctions | Q4_K_M | 5.8 GB | 21.5 t/s | 33K | B62, Decent | |
| Q3_K_L | 20.4 GB | 12.4 t/s | 16K | D22, Barely runs | ||
| IQ2_M | 18.2 GB | 20.7 t/s | 33K | B49, Decent | ||
XiYanSQL QwenCoder 32B 2504ChatCode | Q3_K_M | 20.2 GB | 6.5 t/s | 33K | F14, Won't fit | |
Internlm Chat 7BChat | FP16 | 15.4 GB | 7.0 t/s | 2K | C44, Usable | |
Pollux Judge 32BChat | Q2_K | 18.5 GB | 7.3 t/s | 33K | C31, Usable | |
| Q4_K_M | 7.9 GB | 19.4 t/s | 131K | B59, Decent | ||
Llama Guard 3 8BChat | BF16 | 20.2 GB | 6.1 t/s | — | F13, Won't fit | |
Yi 34BChat | IQ4_XS | 19.8 GB | 5.6 t/s | 4K | D16, Barely runs | |
| Q2_K | 19.4 GB | 31.9 t/s | 262K | B49, Decent | ||
Nemotron H 8B Reasoning 128KChatReasoning | BF16 | 20.4 GB | 6.0 t/s | — | F11, Won't fit | |
Yi 34B ChatChat | IQ4_XS | 19.8 GB | 5.6 t/s | 4K | D16, Barely runs | |
Qwen2.5 Coder 32BChatCodeFunctions | IQ3_M | 19.3 GB | 6.9 t/s | 33K | D23, Barely runs | |
Qwen1.5 14BChat | IQ3_M | 20.1 GB | 12.9 t/s | 33K | D24, Barely runs |
Specs
- Chip
- M5
- CPU cores
- 10
- GPU cores
- 10
- Unified memory
- 24 GB
- For models
- 20.5 GB (macOS keeps 3.5 GB)
- Memory bandwidth
- 153.6 GB/s
- FP16 (dense)
- 16 TFLOPS
- Neural Engine
- 38 TOPS
- Released
- 2025-10-22
MacBook Pro 14" M5 FAQ
How much memory is available for AI on the MacBook Pro 14" M5 (24 GB)?
About 20.5 GB of the 24 GB. macOS keeps roughly 3.5 GB for itself and other apps.
What is the largest model the MacBook Pro 14" M5 (24 GB) can run?
Qwen3.6 35B A3B (36.0B) at IQ4_XS uses 19.8 GB, 97% of the available memory, and runs at about 31.1 tok/s.
Can the MacBook Pro 14" M5 (24 GB) run Gemma 4 26B A4B IT?
Yes. Gemma 4 26B A4B IT fits at Q4_K_M, using 16.5 GB of the 20.5 GB available, and it is the best model that runs well on this machine.
Can the MacBook Pro 14" M5 (24 GB) run 70B models?
No. Llama 3.3 70B Instruct needs 51.3 GB at Q4, more than the 20.5 GB available.
How long does the MacBook Pro 14" M5 (24 GB) take to read a long prompt?
On Gemma 4 26B A4B IT, about 11 s for a 4K-token prompt and 1.5 min for a 32K-token document before the first word appears.
Is renting a GPU cheaper than buying the MacBook Pro 14" M5?
An RTX 3090 (24 GB) rents from $0.17/hr.