Hardware
Apple M5 Pro
Apple M5 Pro has 24 / 48 / 64 GB Unified and 307 GB/s. See which open models fit, at which quant, and roughly how fast. Figures below use the default memory (24 GB). Updated 2026-10-24.
Memory24 / 48 / 64 GB Unified
Bandwidth307 GB/s
Kindmac
Year2026
Apple March 2026: up to 307 GB/s and 64GB. Source.
RSS for new models on this device
Models that fit
| Model | Verdict | Quant | Context | tok/s |
|---|---|---|---|---|
| DeepSeek R1 Distill Llama 70B 70.6B | Tight fit | UD-IQ1_S | 4K | 9.8–13 |
| Llama 3.3 70B Instruct 70.6B | Tight fit | UD-IQ1_S | 4K | 9.8–13 |
| Hermes 4 70B 70.6B | Tight fit | UD-IQ1_S | 4K | 9.8–13 |
| GLM 4.7 Flash 29.9B | Tight fit | UD-Q4_K_XL | 8K | 41–55 |
| Qwen3 Coder 30B A3B Instruct 30.5B | Tight fit | UD-Q4_K_XL | 4K | 79–108 |
| Qwen3 30B A3B Instruct 2507 30.5B | Tight fit | UD-Q4_K_XL | 4K | 79–108 |
| Qwen3 VL 30B A3B Instruct 30.5B | Tight fit | UD-Q4_K_XL | 4K | 79–107 |
| Qwen3 30B A3B Thinking 2507 30.5B | Tight fit | UD-Q4_K_XL | 4K | 79–107 |
| Qwen3 32B 32.8B | Tight fit | UD-Q3_K_XL | 4K | 9.7–13 |
| Qwen3 VL 32B Instruct 32.8B | Tight fit | UD-Q3_K_XL | 4K | 9.7–13 |
| Qwen3.5 35B A3B 34.7B | Tight fit | UD-IQ4_XS | 8K | 100–137 |
| Qwen3.6 35B A3B 34.7B | Tight fit | UD-IQ4_XS | 8K | 99–135 |
| Tiel Coder 35B A3B 34.7B | Tight fit | UD-IQ4_XS | 8K | 99–135 |
| Nemotron 3 Nano 30B A3B 31.6B | Tight fit | IQ4_NL | 4K | 87–118 |
| LFM2.5 24B A2B 24.0B | Tight fit | UD-Q5_K_XL | 4K | 92–126 |
| gemma 4 26B A4B it 25.2B | Tight fit | UD-Q4_K_XL | 8K | 36–49 |
| GLM 4.7 Flash REAP 23B A3B 23.0B | Tight fit | UD-Q5_K_XL | 8K | 64–88 |
| Kwaipilot KAT Coder V2.5 Dev 34.7B | Tight fit | Q3_K_L | 8K | 52–71 |
| bottlecapai ThinkingCap gemma 4 26B A4B it 25.2B | Tight fit | Q4_K_M | 4K | 44–60 |
| gpt oss 20b 20.9B | Runs | UD-Q8_K_XL | 8K | 42–57 |
| Qwen2.5 14B Instruct 14.8B | Tight fit | Q8_0 | 8K | 9.8–13 |
| Qwen3 14B 14.8B | Tight fit | Q8_0 | 8K | 9.9–14 |
| NousResearch Hermes 4 14B 14.8B | Tight fit | Q8_0 | 8K | 9.9–14 |
| DeepSeek R1 Distill Qwen 14B 14.8B | Tight fit | Q8_0 | 8K | 9.8–13 |
| Qwen2.5 Coder 14B Instruct 128K 14.8B | Tight fit | Q8_0 | 8K | 9.8–13 |
| Cyber Tiel Coder 35B A3B GGUF MTP 35.5B | Tight fit | Q3_K_M (est.) | 8K | 104–141 |
| Qwen3.8 35B A3B Distill 35.5B | Tight fit | Q3_K_M | 8K | 102–139 |
| phi 4 14.7B | Tight fit | Q8_0 | 8K | 9.8–13 |
| gemma 4 12b it 11.9B | Runs | UD-Q8_K_XL | 8K | 10–14 |
| Qwen3.6 14B A3B FableVibes 13.8B | Runs | Q8_0 | 8K | 50–68 |
| GLM 4.6V Flash 9.4B | Runs | UD-Q8_K_XL | 8K | 14–19 |
| GLM Z1 9B 0414 9.4B | Runs | UD-Q8_K_XL | 8K | 14–19 |
| GLM 4 9B 0414 9.4B | Runs | UD-Q8_K_XL | 8K | 14–19 |
| Qwen3.8 9B Distill 9.2B | Runs great | Q8_0 | 8K | 17–23 |
| Qwen3 VL 8B Instruct 8.2B | Tight fit | BF16 | 8K | 9.6–13 |
| DeepSeek R1 0528 Qwen3 8B 8.2B | Tight fit | BF16 | 8K | 9.6–13 |
| Qwen3 8B 8.2B | Tight fit | BF16 | 8K | 9.6–13 |
| Qwen3 VL 8B Thinking 8.2B | Tight fit | BF16 | 8K | 9.6–13 |
| DeepSeek R1 Distill Llama 8B 8.0B | Tight fit | BF16 | 8K | 9.8–13 |
| Llama 3.1 8B Instruct 8.0B | Tight fit | BF16 | 8K | 9.8–13 |
105 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 Pro at UD-IQ1_S, 4K context, about 9.8–13 tok/s. Weights 14.8 GB plus 1.25 GB of KV cache — 16.5 GB of 18.0 GB usable. Weights shown in the verdict use GiB (the same 1024³ counting as GPU memory).