RunsLocalGPU checker Local AI toolsv2.2 · data Oct 17, 2026

What AI models can the M5 Pro 24GB run?

With 24 GB unified memory, the M5 Pro 24GB runs 10 of 25 popular open models fully in fast memory at Q4_K_M or better (8K context). The largest that fit are Gemma 4 26B A4B (Q4_K_M), Mistral Small 3.1 24B (Q4_K_M), gpt-oss-20b (Q5_K_M). Fastest: Gemma 4 26B A4B at about 80 tokens/sec and Gemma 4 E2B at about 75 tokens/sec.

Check with your own RAM and context length

Model	Result	Best quant	Memory needed	Est. speed
Gemma 4 26B A4B MoE · vision	Runs on GPU	Q4_K_M	16.4 GB	80 tok/s
Gemma 4 E2B small	Runs on GPU	Q8_0	6.1 GB	75 tok/s
gpt-oss-20b MoE · reasoning	Runs on GPU	Q5_K_M	15.8 GB	72 tok/s
Mellum2.1 12B-A2.5B Thinking MoE · coding	Runs on GPU	Q8_0	13.3 GB	69 tok/s
Ministral 3 3B small	Runs on GPU	Q8_0	4.8 GB	46 tok/s
Gemma 4 E4B small · audio	Runs on GPU	Q8_0	9.3 GB	39 tok/s
Qwen3.5 9B general	Runs on GPU	Q8_0	11.3 GB	18 tok/s
Phi-4 14B reasoning	Runs on GPU	Q6_K	14.2 GB	15 tok/s
Gemma 4 12B general · audio	Runs on GPU	Q8_0	13.8 GB	14 tok/s
Mistral Small 3.1 24B general	Runs on GPU	Q4_K_M	16.4 GB	13 tok/s
Qwen3.6 35B-A3B MoE · coding	Lower quality fit	IQ2_M	12.8 GB	150+ tok/s
Nemotron 3.5 Lightning 30B-A3B MoE · agentic	Lower quality fit	Q3_K_M	15.5 GB	126 tok/s
Hermes 4.3 36B general	Lower quality fit	IQ2_M	14.2 GB	15 tok/s
Qwen3.6 27B coding	Lower quality fit	Q3_K_M	14.3 GB	14 tok/s
Qwen3.8 27B coding	Lower quality fit	Q3_K_M	14.7 GB	14 tok/s
Granite 4.2 30B general	Lower quality fit	Q3_K_M	15.1 GB	13 tok/s
Gemma 4 31B general · vision	Lower quality fit	Q3_K_M	16.2 GB	12 tok/s
Llama 3.3 70B general	Too big	Q4_K_M	45.9 GB	—
Llama 4 Scout MoE · vision	Too big	Q4_K_M	68.1 GB	—
gpt-oss-120b MoE · reasoning	Too big	Q4_K_M	72.0 GB	—
Qwen3.5 122B-A10B MoE	Too big	Q4_K_M	75.3 GB	—
Qwen3.8-Flash-Next MoE	Too big	Q4_K_M	107.8 GB	—
GLM-5.3-Flash MoE · coding	Too big	Q4_K_M	196.0 GB	—
DeepSeek V4.1-Flash MoE · coding	Too big	Q4_K_M	335.8 GB	—
Muse Spark Open 816B-A42B MoE · vision	Too big	Q4_K_M	497.3 GB	—

Specs used: 24 GB unified memory, 307 GB/s memory bandwidth. Speeds are estimates for generating text with llama.cpp-style runtimes; real numbers usually land within about 20%.

Apps that support the M5 Pro 24GB

Ollama LM Studio Jan GPT4All AnythingLLM llama.cpp MLX LM KoboldCpp Text Generation WebUI LocalAI