Hardware

Apple M5 Max (32-core GPU)

Apple M5 Max (32-core GPU) has 36 GB Unified and 460 GB/s. See which open models fit, at which quant, and roughly how fast. Figures below use the default memory (36 GB). Updated 2026-10-17.

Memory36 GB Unified

Bandwidth460 GB/s

Kindmac

Year2026

Apple MacBook Pro specs: 32-core GPU at 460 GB/s, 36GB. Source.

RSS for new models on this device

Models that fit

Model	Verdict	Quant	Context	tok/s
Qwen2.5 32B Instruct 32.8B	Tight fit	Q5_K_M	8K	10.0–14
Qwen3 32B 32.8B	Tight fit	UD-Q5_K_XL	8K	10.0–14
Qwen2.5 Coder 32B Instruct 32.8B	Tight fit	Q5_K_M	8K	10.0–14
DeepSeek R1 Distill Qwen 32B 32.8B	Tight fit	Q5_K_M	8K	10.0–14
Qwen3 VL 32B Instruct 32.8B	Tight fit	UD-Q5_K_XL	8K	10.0–14
Qwen3.8 27B 27.3B	Tight fit	UD-Q6_K_XL	8K	9.8–13
Qwen3.6 27B 26.9B	Tight fit	UD-Q6_K_XL	8K	9.7–13
Qwen3.5 27B 26.9B	Tight fit	UD-Q6_K_XL	8K	9.7–13
Kwaipilot KAT Coder V2.5 Dev 34.7B	Tight fit	Q5_K_M	8K	54–73
Qwen3.8 35B A3B Distill 35.5B	Tight fit	Q5_K_M	8K	110–149
Qwen3.5 35B A3B 34.7B	Tight fit	UD-Q5_K_XL	8K	103–141
Qwen3.6 35B A3B 34.7B	Tight fit	UD-Q5_K_XL	8K	102–140
NVIDIA Nemotron 3.5 Lightning 30B A3B 32.9B	Tight fit	UD-Q5_K_S	8K	90–122
Devstral Small 2 24B Instruct 2512 23.6B	Tight fit	Q8_0	8K	9.6–13
Mistral Small 3.2 24B Instruct 2506 23.6B	Tight fit	Q8_0	8K	9.6–13
Magistral Small 2509 23.6B	Tight fit	Q8_0	8K	9.6–13
NVIDIA Nemotron 3 Nano Omni 30B A3B Reasoning 31.6B	Tight fit	UD-Q5_K_S	8K	91–124
Nemotron 3 Nano 30B A3B 31.6B	Tight fit	UD-Q5_K_XL	4K	89–122
GLM 4.7 Flash 29.9B	Tight fit	UD-Q6_K_XL	8K	42–57
Qwen3 Coder 30B A3B Instruct 30.5B	Tight fit	UD-Q6_K_XL	8K	75–102
Qwen3 30B A3B Instruct 2507 30.5B	Tight fit	UD-Q6_K_XL	8K	75–102
Qwen3 VL 30B A3B Instruct 30.5B	Tight fit	UD-Q6_K_XL	8K	75–102
Qwen3 30B A3B Thinking 2507 30.5B	Tight fit	UD-Q6_K_XL	8K	75–102
Qwen3 Coder Next 79.7B	Tight fit	UD-Q2_K_XL	8K	47–65
Llama 3.3 70B Instruct 70.6B	Runs	UD-IQ2_XXS	8K	11–16
DeepSeek R1 Distill Llama 70B 70.6B	Runs	UD-IQ2_XXS	8K	11–16
Hermes 4 70B 70.6B	Runs	UD-IQ2_XXS	8K	11–16
gemma 4 26B A4B it 25.2B	Tight fit	Q8_0	8K	40–55
LFM2.5 24B A2B 24.0B	Tight fit	Q8_0	8K	90–123
Qwen3 Next 80B A3B Instruct 79.7B	Tight fit	UD-IQ2_XXS	8K	141–192
Qwen3 Next 80B A3B Thinking 79.7B	Tight fit	UD-IQ2_XXS	8K	141–192
GLM 4.7 Flash REAP 23B A3B 23.0B	Tight fit	UD-Q8_K_XL	4K	66–90
gpt oss 20b 20.9B	Runs great	UD-Q8_K_XL	8K	63–86
Qwen2.5 14B Instruct 14.8B	Runs	Q8_0	8K	15–20
NousResearch Hermes 4 14B 14.8B	Runs	Q8_0	8K	15–20
DeepSeek R1 Distill Qwen 14B 14.8B	Runs	Q8_0	8K	15–20
Qwen2.5 Coder 14B Instruct 128K 14.8B	Runs	Q8_0	8K	15–20
Qwen3 14B 14.8B	Runs	UD-Q8_K_XL	8K	13–17
phi 4 14.7B	Runs	Q8_0	8K	15–20
GLM 4.6V Flash 9.4B	Runs	BF16	8K	13–18

90 more models in the catalog do not fit this default profile, or ranked below the first 40 that do.

Example: Tight fit fully in unified memory on your Apple M5 Max (32-core GPU) at Q5_K_M, 8K context, about 10.0–14 tok/s. Weights 21.7 GB plus 2.00 GB of KV cache — 24.2 GB of 27.0 GB usable. Up to 16K context still fits. Weights shown in the verdict use GiB (the same 1024³ counting as GPU memory).