CanItRun Logocanitrun.

Apple M1

The Apple M1 ships in 8–16 GB unified-memory configurations at 68 GB/s. Across those configurations it runs 26 of our 84 tracked models natively in VRAM at 8k context.

More memory means more of our tracked models fit natively — see which configuration you need below.

Jump to:16 GB8 GB
ConfigurationBandwidthCPU coresNative models+ Offload
16 GB68 GB/s8 (4P + 4E)26 / 840
8 GB68 GB/s8 (4P + 4E)0 / 840
Vendor: Apple
Memory type: LPDDR4X
Compute backend: METAL
Software: MLX gives the best performance on Apple Silicon; llama.cpp Metal backend is a solid alternative. Both are well-supported by Ollama.

How much of the Apple M1's memory is actually usable?

macOS and background apps need a slice of the pool before a model gets to use it — this site reserves 8GB on every unified-memory GPU, the same baseline used everywhere else on this site. What's left is real headroom for a model's weights and KV cache:

macOS + background apps (8 GB reserved)usable for model weights + KV cacheVertical lines mark roughly where a 7B/14B/32B/... dense model at Q4_K_M lands.

Apple M1 (16GB)

With 16 GB LPDDR4X at 68 GB/s, this configuration runs 26 models natively. It handles smaller models (7B–14B) at Q4–Q5 quantization.

Apple M1 (16GB): launched November 2020 as the top memory tier of the original Apple Silicon chip, doubling the base 8 GB configuration for around $200 across the MacBook Air, 13-inch MacBook Pro, and Mac mini. Same 8-core CPU (4P+4E) and 68 GB/s LPDDR4X bandwidth as the 8GB model — only the memory pool grew.

With this site's standard 8 GB unified-memory reservation applied, this configuration has 8 GB of real headroom — enough to clear the floor the 8GB model can't reach at all. This site's calculator fits Llama 3.1 8B at Q5_K_M (7.58 GB, ~8 tok/s) and Qwen2.5 7B at Q6_K (7.51 GB, ~8.1 tok/s). A few 12-14B models also fit, but only at the bottom of the quant ladder — Qwen3 14B needs Q2_K and Gemma 3 12B needs Q3_K_M, both trading real quality for capacity. 26 of the 84 tracked models fit natively; nothing above 14B parameters does.

Community llama.cpp benchmarks on the base M1 chip report roughly 14 tok/s decode on a 7B model at Q4_0 and 108 tok/s prompt processing at F16 (ggml-org/llama.cpp Discussion #4167) — figures set by the 7-8 core GPU and 68 GB/s bandwidth, independent of installed memory size. That 7B figure runs faster than the ~8 tok/s this site projects for the larger 8-14B models this 16 GB pool can just barely hold, a reminder that model size drives real-world speed as much as chip generation does.

Models the 16 GB configuration runs natively (26)

Show 21 more

Apple M1 (8GB)

With 8 GB LPDDR4X at 68 GB/s, this configuration runs 0 models natively. It's best for smaller models under 8B parameters.

Apple M1 (8GB): the base configuration of Apple's first Apple Silicon Mac, launched November 2020 on the entry-level MacBook Air, 13-inch MacBook Pro, and Mac mini. Same 8-core CPU (4P+4E) and 68 GB/s LPDDR4X bandwidth as the 16GB model — Apple never sold a smaller Apple Silicon Mac, so 8 GB is the practical floor for local inference on any Mac.

This site reserves 8 GB of unified memory for macOS and background apps on every Apple Silicon GPU — the same baseline used everywhere on this site — which leaves this 8 GB machine with 0 GB of real headroom for a model's weights and KV cache. None of the 84 models tracked on this site fit natively in memory on this configuration; even the most aggressively quantized small models need more room than the reservation leaves free.

MLX and llama.cpp's Metal backend both run on this chip, but neither changes the arithmetic: an 8 GB pool minus an 8 GB OS reservation is 0 GB free, regardless of software. Anyone shopping specifically for local LLM inference should look at the 16GB M1 at minimum, or an M1 Pro for real headroom.

Models the 8 GB configuration runs natively (0)

None.

Too large for any Apple M1 configuration (58)

Continue reading

tutorials10 min
How to Run LLMs Locally on Mac (M1–M4 Guide)
hardware11 min
Apple Silicon for LLMs: M1-M5 Complete Guide
tutorials11 min
Getting Started with Local LLMs (2026): The Complete Beginner's Guide

Frequently asked questions

How much memory does the Apple M1 have?
The Apple M1 ships in 2 unified-memory configurations: 16 GB and 8 GB, all at 68 GB/s.
Should I get the 8 GB or 16 GB Apple M1?
The 8 GB configuration doesn't run any of the models tracked on this site natively in VRAM — after this site's standard OS-memory reservation, it has no real headroom left for a model's weights. The 16 GB configuration is the practical minimum for local LLM inference on this chip, fitting Bonsai 27B, Qwen3 14B, Phi-4 14B Instruct, and 23 more models natively.
How much VRAM does the Apple M1 (16GB) have?
The Apple M1 (16GB) has 16 GB of LPDDR4X with 68 GB/s memory bandwidth (unified system memory, shared between CPU and GPU).
What is the Apple M1 (16GB) best for?
With 16 GB of VRAM, the Apple M1 (16GB) handles smaller models (7B–14B) at Q4–Q5 quantization — ideal for entry-level local LLM experimentation and lightweight inference.
What LLMs can the Apple M1 (16GB) run locally?
The Apple M1 (16GB) can run 26 of the 84 open-weight models tracked by CanItRun natively in VRAM at 8k context. Top options include: Llama 3.1 8B Instruct at Q5_K_M, Llama 3.2 3B Instruct at Q8_0, Llama 3.2 1B Instruct at FP32.
Can the Apple M1 (16GB) run Llama 3.3 70B Instruct?
The Apple M1 (16GB) does not have enough VRAM to run Llama 3.3 70B Instruct. You would need more VRAM or a lower quantization level.
Show 8 more questions
Can the Apple M1 (16GB) run Qwen 3.6 27B?
The Apple M1 (16GB) does not have enough VRAM to run Qwen 3.6 27B. You would need more VRAM or a lower quantization level.
Can the Apple M1 (16GB) run Llama 3.1 8B Instruct?
Yes. The Apple M1 (16GB) runs Llama 3.1 8B Instruct natively in VRAM at Q5_K_M quantization, achieving approximately 8 tokens per second.
How much VRAM does the Apple M1 (8GB) have?
The Apple M1 (8GB) has 8 GB of LPDDR4X with 68 GB/s memory bandwidth (unified system memory, shared between CPU and GPU).
What is the Apple M1 (8GB) best for?
With 8 GB of VRAM, the Apple M1 (8GB) is best for running compact models (1B–8B) at low quantization, suitable for edge inference, prototyping, and lightweight tasks.
What LLMs can the Apple M1 (8GB) run locally?
The Apple M1 (8GB) cannot run any of the 84 tracked models fully in VRAM at 8k context. It may handle smaller models with CPU offload.
Can the Apple M1 (8GB) run Llama 3.3 70B Instruct?
The Apple M1 (8GB) does not have enough VRAM to run Llama 3.3 70B Instruct. You would need more VRAM or a lower quantization level.
Can the Apple M1 (8GB) run Qwen 3.6 27B?
The Apple M1 (8GB) does not have enough VRAM to run Qwen 3.6 27B. You would need more VRAM or a lower quantization level.
Can the Apple M1 (8GB) run Llama 3.1 8B Instruct?
The Apple M1 (8GB) does not have enough VRAM to run Llama 3.1 8B Instruct. You would need more VRAM or a lower quantization level.