Hardware

Apple M4 Max (32-core GPU)

Apple M4 Max (32-core GPU) has 36 GB Unified and 410 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

Bandwidth410 GB/s

Kindmac

Year2024

Apple MacBook Pro specs: 14-core CPU / 32-core GPU at 410 GB/s, 36GB. Source.

RSS for new models on this device

Models that fit

Model	Verdict	Quant	Context	tok/s
Kwaipilot KAT Coder V2.5 Dev 34.7B	Tight fit	Q5_K_M	8K	48–65
Qwen3.8 35B A3B Distill 35.5B	Tight fit	Q5_K_M	8K	98–133
Qwen3.5 35B A3B 34.7B	Tight fit	UD-Q5_K_XL	8K	92–125
Qwen3.6 35B A3B 34.7B	Tight fit	UD-Q5_K_XL	8K	91–125
NVIDIA Nemotron 3.5 Lightning 30B A3B 32.9B	Tight fit	UD-Q5_K_S	8K	80–109
NVIDIA Nemotron 3 Nano Omni 30B A3B Reasoning 31.6B	Tight fit	UD-Q5_K_S	8K	81–110
Nemotron 3 Nano 30B A3B 31.6B	Tight fit	UD-Q5_K_XL	4K	80–109
GLM 4.7 Flash 29.9B	Tight fit	UD-Q6_K_XL	8K	38–51
Qwen3 Coder 30B A3B Instruct 30.5B	Tight fit	UD-Q6_K_XL	8K	66–91
Qwen3 30B A3B Instruct 2507 30.5B	Tight fit	UD-Q6_K_XL	8K	66–91
Qwen3 VL 30B A3B Instruct 30.5B	Tight fit	UD-Q6_K_XL	8K	66–91
Qwen3 30B A3B Thinking 2507 30.5B	Tight fit	UD-Q6_K_XL	8K	66–91
Qwen3 Coder Next 79.7B	Tight fit	UD-Q2_K_XL	8K	42–58
Llama 3.3 70B Instruct 70.6B	Runs	UD-IQ2_XXS	8K	10–14
DeepSeek R1 Distill Llama 70B 70.6B	Runs	UD-IQ2_XXS	8K	10–14
Hermes 4 70B 70.6B	Runs	UD-IQ2_XXS	8K	10–14
gemma 4 26B A4B it 25.2B	Tight fit	Q8_0	8K	36–49
LFM2.5 24B A2B 24.0B	Tight fit	Q8_0	8K	81–110
Qwen3 Next 80B A3B Instruct 79.7B	Tight fit	UD-IQ2_XXS	8K	126–172
Qwen3 Next 80B A3B Thinking 79.7B	Tight fit	UD-IQ2_XXS	8K	126–172
GLM 4.7 Flash REAP 23B A3B 23.0B	Tight fit	UD-Q8_K_XL	4K	59–81
gpt oss 20b 20.9B	Runs great	UD-Q8_K_XL	8K	56–76
Qwen2.5 14B Instruct 14.8B	Runs	Q8_0	8K	13–18
NousResearch Hermes 4 14B 14.8B	Runs	Q8_0	8K	13–18
DeepSeek R1 Distill Qwen 14B 14.8B	Runs	Q8_0	8K	13–18
Qwen2.5 Coder 14B Instruct 128K 14.8B	Runs	Q8_0	8K	13–18
Qwen3 14B 14.8B	Runs	UD-Q8_K_XL	8K	11–15
phi 4 14.7B	Runs	Q8_0	8K	13–18
GLM 4.6V Flash 9.4B	Runs	BF16	8K	12–16
GLM Z1 9B 0414 9.4B	Runs	BF16	8K	12–16
GLM 4 9B 0414 9.4B	Runs	BF16	8K	12–16
GLM 4.1V 9B Thinking 9.4B	Runs	BF16	8K	12–16
Qwen3.5 9B 9.0B	Runs	BF16	8K	12–17
MiMo V2.6 Distill Qwen 9B 9.0B	Runs	BF16	8K	12–17
Ministral 3 8B Instruct 2512 8.5B	Runs	BF16	8K	12–17
Ministral 3 8B Reasoning 2512 8.5B	Runs	BF16	8K	12–17
Qwen3 VL 8B Instruct 8.2B	Runs	BF16	8K	13–17
DeepSeek R1 0528 Qwen3 8B 8.2B	Runs	BF16	8K	13–17
Qwen3 8B 8.2B	Runs	BF16	8K	13–17
Qwen3 VL 8B Thinking 8.2B	Runs	BF16	8K	13–17

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 M4 Max (32-core GPU) at Q5_K_M, 8K context, about 48–65 tok/s. Weights 23.3 GB plus 0.19 GB of KV cache — 24.0 GB of 27.0 GB usable. Up to 128K context still fits. Weights shown in the verdict use GiB (the same 1024³ counting as GPU memory).