LocalLLMChecker

Unified memory · Apple

What AI models can an Apple M4 16GB run?

16 GB covers almost everything people actually use day to day, including the 24B to 32B class that most reviewers consider the practical ceiling for a single consumer card. Memory is shared with the CPU, so the practical ceiling is lower than the sticker number: macOS hands roughly three quarters of it to the GPU by default.

16 GB of unified memory, of which macOS gives the GPU about two thirds. That is a card with roughly 10 GB for models, at 120 GB/s. It runs 7B to 9B models at modest speed and is a fine way to try local AI on a machine you already own.

Memory16 GB
Bandwidth120 GB/s
Usable for a model10.72 GB
Runtime backendMETAL

The short answer

Assuming an 8k context window and default settings, these are the models worth downloading first.

Best all-rounder

Qwen3.5 9B

Q6_K · 8.04 GB · about 11.8 tokens/s

Best for code

Qwen2.5 Coder 7B Instruct

Q8_0 · 8.34 GB · about 11 tokens/s

Largest that still runs well

Qwen3 14B

14.77B parameters · IQ4_XS · about 10 tokens/s

Fastest useful answer

LFM2.5 8B A1B

about 28.7 tokens/s · 8.78 GB

See how fast it feels

Apple M4 16GB running Qwen3.5 9B at Q6_K

Wait for the first word7.1 s
Then writes at11.8 tok/s
Whole answer22 s

YouWhy does my model use more memory when the conversation gets longer?

Model

Simulated from our estimate at a 512-token question, not a recording. Assumes nothing else is competing for the GPU.

Where it sits

The GPU can use about 10.72 GB of this machine's 16 GB by default, so for planning it behaves like a 8 GB card with different bandwidth. The 8 GB guide explains what that tier runs and where it runs out, and the bandwidth guide explains why 120 GB/s decides how fast it answers.

Our indicative price for this machine is $599 new, reviewed 2026-09-01. That is 2.7 GB of memory and 20 GB/s of bandwidth for every $100. Prices move, so treat these as a place to start a search and not a quote.

Can an Apple M4 16GB run it?

A direct answer for the models people ask about most, each with the settings we would use.

Qwen2.5 7B InstructLlama 3.1 8B InstructQwen3 8BQwen3.5 9BGemma 4 12B InstructQwen3 14BPhi 4GPT OSS 20BQwen3.5 27BGemma 4 31B InstructQwen3 32BQwen3 30B A3B

Every model, scored on this device

Each row uses the highest-quality quantization that both fits and stays conversational. Speed is a single-stream estimate at 8k context.

Model Params Verdict Download Memory used Speed Max context
Qwen3.5 9B
Qwen 		9.65B Runs well Q6_K 8.04 GB 11.8 t/s 64k
Gemma 4 12B Instruct
Gemma 		11.96B Runs well Q4_K_M 8.41 GB 10.2 t/s 16k
Qwen3 8B
Qwen 		8.19B Runs well Q6_K 7.76 GB 11.8 t/s 16k
Gemma 4 E4B Instruct
Gemma 		8B Runs well Q8_0 8.39 GB 10.1 t/s 128k
Qwen3.5 4B
Qwen 		4.66B Runs well Q8_0 5.24 GB 16.6 t/s 128k
Gemma 4 E2B Instruct
Gemma 		5.12B Runs well Q8_0 5.41 GB 15.9 t/s 128k
Granite 4.1 8B
Granite 		8.79B Runs well Q6_K 8.34 GB 10.2 t/s 16k
Qwen3 14B
Qwen 		14.77B Fits, but slow IQ4_XS 8.97 GB 10 t/s 16k
Qwen2.5 VL 7B Instruct
Qwen 		8.29B Runs well Q8_0 9 GB 10.1 t/s 32k
Llama 3.1 8B Instruct
Llama 		8.03B Runs well Q6_K 7.51 GB 12.1 t/s 32k
LFM2.5 8B A1B
Liquid 		8.47B (1.57B active) Runs great Q8_0 8.78 GB 28.7 t/s 64k
Qwen2.5 7B Instruct
Qwen 		7.62B Runs well Q8_0 8.34 GB 11 t/s 32k
Gemma 3 12B Instruct
Gemma 		12.19B Runs well Q4_K_M 8.04 GB 10.6 t/s 32k
Qwen2.5 Coder 7B Instruct
Qwen 		7.62B Runs well Q8_0 8.34 GB 11 t/s 32k
Internlm3 8B Instruct
InternLM 		8.8B Runs well Q6_K 7.47 GB 12.2 t/s 32k
Mistral Nemo Instruct 2407
Mistral 		12.25B Runs well Q4_K_M 8.55 GB 10.5 t/s 16k
Gemma 3 4B Instruct
Gemma 		4.3B Runs well Q8_0 4.88 GB 17.9 t/s 128k
Qwen2.5 14B Instruct
Qwen 		14.77B Runs well Q3_K_M 8.63 GB 10.4 t/s 16k
OLMo 3 7B Instruct
OLMo 		7.3B Runs well Q6_K 8.45 GB 10.5 t/s 16k
Qwen2.5 Coder 14B Instruct
Qwen 		14.77B Runs well Q3_K_M 8.63 GB 10.4 t/s 16k
Qwen3 4B
Qwen 		4.02B Runs well Q8_0 5.43 GB 16 t/s 32k
Mistral 7B Instruct V0.3
Mistral 		7.25B Runs well Q8_0 8.55 GB 10.1 t/s 16k
Gemma 2 9B Instruct
Gemma 		9.24B Runs well Q5_K_M 8.45 GB 10.1 t/s 8k
Phi 4
Phi 		14.66B Runs well Q3_K_M 8.65 GB 10.2 t/s 16k
Mistral 7B Instruct V0.2
Mistral 		7.24B Runs well Q8_0 8.54 GB 10.2 t/s 16k
Granite 4.1 3B
Granite 		3.4B Runs well Q8_0 4.31 GB 20.5 t/s 64k
Qwen2.5 3B Instruct
Qwen 		3.09B Runs well Q8_0 3.65 GB 24.4 t/s 32k
Phi 4 Mini Instruct
Phi 		3.84B Runs well Q8_0 5.14 GB 17 t/s 32k
Llama 3.2 3B Instruct
Llama 		3.21B Runs well Q8_0 4.39 GB 20.1 t/s 64k
Phi 3 Mini 4k Instruct
Phi 		3.82B Runs well Q8_0 4.87 GB 18.4 t/s 4k
Show the other 56 models
DeepSeek Coder 7B Instruct V1.5 6.91B Runs well Q5_K_M 8.7 GB 10.1 t/s 4k
LFM2.5 2.6B 2.7B Runs great Q8_0 3.11 GB 29.2 t/s 128k
Qwen3.5 2B 2.27B Runs great Q8_0 2.67 GB 34.6 t/s 256k
DeepSeek Coder 6.7B Instruct 6.74B Runs well Q4_K_M 8.17 GB 10.6 t/s 8k
Gemma 2 2B Instruct 2.61B Runs great Q8_0 3.51 GB 25.6 t/s 8k
Starcoder2 3B 3.03B Runs great Q8_0 3.46 GB 27.5 t/s 16k
Qwen3 1.7B 2.03B Runs great Q8_0 3.19 GB 28.3 t/s 32k
DeepSeek R1 Distill Qwen 1.5B 1.78B Runs great Q8_0 2.27 GB 46.6 t/s 128k
Qwen2.5 1.5B Instruct 1.54B Runs great Q8_0 2.03 GB 46.8 t/s 32k
SmolLM2 1.7B 1.71B Runs great Q8_0 3.5 GB 25.6 t/s 8k
OLMo 2 0425 1B 1.48B Runs great Q8_0 2.77 GB 36.1 t/s 4k
Llama 3.2 1B Instruct 1.24B Runs great Q8_0 1.79 GB 55.2 t/s 128k
LFM2.5 1.2B Instruct 1.17B Runs great Q8_0 1.56 GB 72.9 t/s 64k
Gemma 3 1B Instruct 1B Runs great Q8_0 1.32 GB 78.5 t/s 32k
Qwen3.5 0.8B 0.87B Runs great Q8_0 1.25 GB 83.9 t/s 256k
Qwen3 0.6B 0.75B Runs great Q8_0 1.89 GB 50.4 t/s 32k
Qwen2.5 0.5B Instruct 0.49B Runs great Q8_0 0.85 GB 141 t/s 32k
LFM2.5 350M 0.35B Runs great Q8_0 0.71 GB 218.3 t/s 64k
Qwen3.5 35B A3B 35.95B Fits only at 2 to 3 bits n/a 13.33 GB 38.3 t/s n/a
GLM 4.7 Flash 31.22B Fits only at 2 to 3 bits n/a 11.84 GB 27.8 t/s n/a
Qwen3 30B A3B 30.53B Fits only at 2 to 3 bits n/a 11.93 GB 24.5 t/s n/a
Qwen3 Coder 30B A3B Instruct 30.53B Fits only at 2 to 3 bits n/a 11.93 GB 24.5 t/s n/a
DeepSeek Coder V2 Lite Instruct 15.71B Just fits n/a 12.54 GB 20.6 t/s n/a
GPT OSS 20B 20.91B Just fits n/a 11.81 GB 19.8 t/s n/a
Phi 3.5 MoE Instruct 41.87B Fits only at 2 to 3 bits n/a 11.42 GB 18.3 t/s n/a
Gemma 4 26B A4B Instruct 25.81B Just fits n/a 13.93 GB 17.7 t/s n/a
Mixtral 8x7B Instruct V0.1 46.7B Fits only at 2 to 3 bits n/a 12.57 GB 11.6 t/s n/a
Gemma 4 31B Instruct 31.27B Fits only at 2 to 3 bits n/a 11.2 GB 7.6 t/s n/a
Granite 4.1 30B 28.87B Fits only at 2 to 3 bits n/a 12.66 GB 6.6 t/s n/a
Mistral Small 24B Instruct 2501 23.57B Just fits n/a 13.32 GB 6.5 t/s n/a
Qwen3.5 27B 27.78B Just fits n/a 13.7 GB 6.4 t/s n/a
Qwen3 32B 32.76B Fits only at 2 to 3 bits n/a 14.08 GB 6.1 t/s n/a
Qwen2.5 32B Instruct 32.76B Fits only at 2 to 3 bits n/a 14.08 GB 6.1 t/s n/a
Qwen2.5 Coder 32B Instruct 32.76B Fits only at 2 to 3 bits n/a 14.08 GB 6.1 t/s n/a
Gemma 3 27B Instruct 27.43B Just fits n/a 13.99 GB 6 t/s n/a
Codestral 22B V0.1 22.25B Just fits n/a 14.7 GB 5.8 t/s n/a
Kimi K2.6 1026.88B Will not run n/a 578.41 GB n/a n/a
Kimi K2 Instruct 1026.41B Will not run n/a 578.15 GB n/a n/a
GLM 5.1 753.86B Will not run n/a 425.01 GB n/a n/a
GLM 5.2 753.33B Will not run n/a 424.72 GB n/a n/a
DeepSeek V3.2 685.4B Will not run n/a 386.4 GB n/a n/a
DeepSeek R1 684.53B Will not run n/a 385.92 GB n/a n/a
MiniMax M3 427.04B Will not run n/a 241.5 GB n/a n/a
GLM 4.5 358.34B Will not run n/a 204.77 GB n/a n/a
Qwen3 235B A22B 235.09B Will not run n/a 134.03 GB n/a n/a
MiniMax M2.5 228.7B Will not run n/a 130.88 GB n/a n/a
MiniMax M2.7 228.69B Will not run n/a 130.87 GB n/a n/a
Qwen3.5 122B A10B 125.09B Will not run n/a 71.01 GB n/a n/a
GPT OSS 120B 116.83B Will not run n/a 59.65 GB n/a n/a
GLM 4.5 Air 110.47B Will not run n/a 63.93 GB n/a n/a
Qwen3 Next 80B A3B Instruct 81.32B Will not run n/a 46.29 GB n/a n/a
Qwen3 Coder Next 79.67B Will not run n/a 45.37 GB n/a n/a
Qwen2.5 VL 72B Instruct 73.41B Will not run n/a 44.29 GB n/a n/a
Qwen2.5 72B Instruct 72.71B Will not run n/a 43.9 GB n/a n/a
Llama 3.3 70B Instruct 70.55B Will not run n/a 42.68 GB n/a n/a
Llama 3 3 Nemotron Super 49B V1 49.87B Will not run n/a 48.55 GB n/a n/a

What it will not run

38 of the 86 architectures we track are out of reach here, even at 2-bit precision.

If you want more headroom

The next steps up in memory, in order. More memory changes which models load at all; more bandwidth changes how fast they answer, so the two are worth weighing separately.

DeviceMemoryBandwidthModels that fit
Radeon RX 7900 XT 20 GB 800 GB/s 64 Check price
GeForce RTX 4090 24 GB 1008 GB/s 69 Check price
GeForce RTX 3090 Ti 24 GB 1008 GB/s 69 Check price
Radeon RX 7900 XTX 24 GB 960 GB/s 69 Check price

Price links go to an Amazon search for the model name and are affiliate links: if you buy through one, we earn a commission at no cost to you. We do not take payment for placement, and the ordering above is by memory capacity alone.

Other devices with 16 GB

Same capacity, different speed. Once a model fits, bandwidth is what separates these.

GeForce RTX 5080

960 GB/s · NVIDIA

GeForce RTX 5070 Ti

896 GB/s · NVIDIA

GeForce RTX 5080 Laptop

768 GB/s · NVIDIA

GeForce RTX 4080 SUPER

736 GB/s · NVIDIA

GeForce RTX 4080

717 GB/s · NVIDIA

GeForce RTX 4070 Ti SUPER

672 GB/s · NVIDIA

Questions people ask about an Apple M4 16GB

How much VRAM does an Apple M4 16GB have?

The Apple M4 16GB has 16 GB of unified memory shared with the CPU, of which roughly 10.72 GB is available to a model once the operating system takes its share.

Can an Apple M4 16GB run a 70B model?

Not usefully. A 70B model needs about 45.37 GB at 4-bit and this device offers 10.72 GB.

What is the best local model for an Apple M4 16GB?

Qwen3.5 9B at Q6_K is the strongest all-round fit: 8.04 GB used, around 11.8 tokens per second at 8k context.