Hardware

Apple M4

Apple M4 has 16 / 24 / 32 GB Unified and 120 GB/s. See which open models fit, at which quant, and roughly how fast. Figures below use the default memory (16 GB). Updated 2026-10-17.

Memory16 / 24 / 32 GB Unified

Bandwidth120 GB/s

Kindmac

Year2024

Apple: M4 bandwidth is 120 GB/s, up to 32GB unified memory. Source.

RSS for new models on this device

Models that fit

Model	Verdict	Quant	Context	tok/s
GLM 4.7 Flash REAP 23B A3B 23.0B	Tight fit	Q3_K_S	8K	37–50
Kwaipilot KAT Coder V2.5 Dev 34.7B	Tight fit	IQ2_S	4K	32–43
Qwen3.5 35B A3B 34.7B	Tight fit	UD-IQ2_XXS	8K	61–83
Qwen3.6 35B A3B 34.7B	Tight fit	UD-IQ2_XXS	8K	60–82
LFM2.5 24B A2B 24.0B	Tight fit	UD-IQ3_XXS	8K	47–63
GLM 4.7 Flash 29.9B	Tight fit	UD-IQ2_XXS	8K	25–34
gemma 4 26B A4B it 25.2B	Tight fit	UD-Q2_K_XL	8K	18–24
gemma 4 E4B it qat 7.5B	Runs great	UD-Q4_K_XL	8K	13–18
Qwen3 Coder 30B A3B Instruct 30.5B	Tight fit	UD-IQ2_XXS	8K	36–49
Qwen3 30B A3B Instruct 2507 30.5B	Tight fit	UD-IQ2_XXS	8K	36–49
Qwen3 VL 30B A3B Instruct 30.5B	Tight fit	UD-IQ2_XXS	8K	36–49
Qwen3 30B A3B Thinking 2507 30.5B	Tight fit	UD-IQ2_XXS	8K	36–49
LFM2.5 8B A1B 8.5B	Tight fit	UD-Q8_K_XL	8K	44–60
Ministral 3 3B Instruct 2512 3.4B	Runs great	Q8_0	8K	15–20
granite 4.0 h micro 3.2B	Runs great	BF16	8K	10–14
Qwen2.5 3B Instruct 3.1B	Runs great	F16	8K	10–14
Qwen2.5 VL 3B Instruct 3.1B	Runs great	BF16	8K	10–14
SmolLM3 3B 3.1B	Runs	BF16	8K	9.8–13
LFM2.5 2.6B 2.7B	Runs great	BF16	8K	11–15
LFM2.5 VL 3B 2.7B	Runs great	BF16	8K	11–15
Qwen3.5 2B 1.9B	Runs great	BF16	8K	17–23
DeepSeek R1 Distill Qwen 1.5B 1.8B	Runs great	BF16	8K	17–24
Qwen3 1.7B 1.7B	Runs great	BF16	8K	15–21
Qwen3 VL 2B Instruct 1.7B	Runs great	BF16	8K	15–21
Qwen3 VL 2B Thinking 1.7B	Runs great	BF16	8K	15–21
Qwen2.5 1.5B Instruct 1.5B	Runs great	F16	8K	20–27
Qwen2.5 Coder 1.5B Instruct 1.5B	Runs great	F16	8K	20–27
Llama 3.2 1B Instruct 1.2B	Runs great	BF16	8K	24–33
LFM2.5 1.2B Instruct 1.2B	Runs great	BF16	8K	25–34
gemma 3 1b it 1.0B	Runs great	BF16	8K	30–40
Qwen3.5 0.8B 0.75B	Runs great	BF16	8K	41–56
Qwen3 0.6B 0.60B	Runs great	BF16	8K	31–42
Qwen2.5 0.5B Instruct 0.49B	Runs great	F16	8K	60–82
Qwen2.5 14B Instruct 14.8B	Tight fit	Q5_K_S	4K	6.0–8.1
Qwen3 14B 14.8B	Tight fit	Q5_K_S	4K	6.0–8.2
NousResearch Hermes 4 14B 14.8B	Tight fit	Q5_K_S	4K	6.0–8.2
Ministral 3 14B Instruct 2512 13.5B	Tight fit	UD-Q5_K_XL	8K	6.0–8.2
Ministral 3 14B Reasoning 2512 13.5B	Tight fit	UD-Q5_K_XL	8K	6.0–8.2
DeepSeek R1 Distill Qwen 14B 14.8B	Tight fit	Q4_K_M	8K	6.2–8.5
Qwen2.5 Coder 14B Instruct 128K 14.8B	Tight fit	Q4_K_M	8K	6.2–8.5

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 at Q3_K_S, 8K context, about 37–50 tok/s. Weights 9.59 GB plus 0.41 GB of KV cache — 10.4 GB of 11.0 GB usable. Weights shown in the verdict use GiB (the same 1024³ counting as GPU memory).