CanItRun Logocanitrun.

Apple M2

The Apple M2 ships in 8–24 GB unified-memory configurations at 100 GB/s. Across those configurations it runs 39 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:24 GB16 GB8 GB
ConfigurationBandwidthCPU coresNative models+ Offload
24 GB100 GB/s8 (4P + 4E)39 / 840
16 GB100 GB/s8 (4P + 4E)26 / 840
8 GB100 GB/s8 (4P + 4E)0 / 840
Vendor: Apple
Memory type: LPDDR5
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 M2'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 M2 (24GB)

With 24 GB LPDDR5 at 100 GB/s, this configuration runs 39 models natively. It comfortably runs 7B–32B models at Q4; 70B-class models typically need CPU offload.

Apple M2 (24GB): the top memory configuration of the base M2 chip, launched June 2022 as Apple's first Apple Silicon chip to offer a 24GB build-to-order option — 8 GB more than the original M1 generation ever offered. Same 8-core CPU (4P+4E) and 100 GB/s LPDDR5-6400 bandwidth as the 8GB and 16GB configs; only the memory pool grew.

16 GB of real headroom after this site's standard 8 GB reservation is the first M2-generation configuration to clear 30B-class models: this site's calculator fits Qwen3 32B at Q2_K (15.5 GB, ~5.8 tok/s) and the mixture-of-experts Qwen3 30B-A3B at Q2_K (13.7 GB, ~17.3 tok/s) — both out of reach on the 16GB sibling. Mid-size models get real headroom too: Llama 3.1 8B and Qwen2.5 7B fit at Q8_0 (10.7 GB and 9.6 GB) at roughly 8.4-9.4 tok/s rather than being squeezed into Q5/Q6. 39 of the 84 tracked models fit natively, 13 more than the 16GB configuration.

MLX and llama.cpp's Metal backend both run maturely on this chip. Because macOS's Metal working-set reservation stays roughly fixed regardless of installed memory, it eats a smaller share of this 24 GB pool than it does on the 16GB or 8GB configs, leaving more headroom for the 30B-class fits above.

Models the 24 GB configuration runs natively (39)

Show 34 more

Apple M2 (16GB)

With 16 GB LPDDR5 at 100 GB/s, this configuration runs 26 models natively. It handles smaller models (7B–14B) at Q4–Q5 quantization.

Apple M2 (16GB): the standard memory configuration of the base M2 chip, launched June 2022 on the redesigned MacBook Air alongside the 13-inch MacBook Pro. Same 8-core CPU (4P+4E) as the 8GB and 24GB configs; 100 GB/s of LPDDR5-6400 bandwidth is a 47% jump over the original M1's 68 GB/s on the same 128-bit bus width.

8 GB of real headroom after this site's standard 8 GB reservation — enough to clear the floor the 8GB config can't reach at all. This site's calculator fits Llama 3.1 8B at Q5_K_M (7.58 GB, ~11.8 tok/s) and Qwen2.5 7B at Q6_K (7.51 GB, ~11.9 tok/s). A few 12-14B models fit too, but only at the bottom of the quant ladder: Qwen3 14B needs Q2_K (7.82 GB). 26 of the 84 tracked models fit natively; nothing above 14B parameters does.

Community llama.cpp benchmarks on the 10-core-GPU base M2 report roughly 22 tok/s decode on a 7B model at Q4_0 and ~181 tok/s prompt processing at Q8_0 (ggml-org/llama.cpp Discussion #4167) — figures set by GPU core count and 100 GB/s bandwidth, independent of installed memory size. Those Q4_0 numbers run faster than the Q5/Q6 figures this site projects above, since a lower-bit quant reads fewer bytes per token at the same bandwidth.

Models the 16 GB configuration runs natively (26)

Show 21 more

Apple M2 (8GB)

With 8 GB LPDDR5 at 100 GB/s, this configuration runs 0 models natively. It's best for smaller models under 8B parameters.

Apple M2 (8GB): the base configuration of Apple's second-generation Apple Silicon chip, launched June 2022 on the entry MacBook Air and 13-inch MacBook Pro. Same 8-core CPU (4P+4E) and 100 GB/s LPDDR5-6400 bandwidth as the 16GB and 24GB configs — Apple's cheapest M2 Mac, and the practical floor for this chip generation.

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 or the extra bandwidth over an equivalent 8GB M1. Anyone shopping specifically for local LLM inference should look at the 16GB M2 at minimum, or an M2 Pro for real headroom.

Models the 8 GB configuration runs natively (0)

None.

Too large for any Apple M2 configuration (45)

Explore

Browse all guides →Browse all models →Browse all GPUs →

Frequently asked questions

How much memory does the Apple M2 have?
The Apple M2 ships in 3 unified-memory configurations: 24 GB, 16 GB, 8 GB, all at 100 GB/s.
Should I get the 8 GB or 24 GB Apple M2?
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 24 GB configuration is the practical minimum for local LLM inference on this chip, fitting Qwen 3.5 35B-A3B (MoE), Qwen3 32B, Nemotron 3 Nano 30B, and 36 more models natively.
How much VRAM does the Apple M2 (24GB) have?
The Apple M2 (24GB) has 24 GB of LPDDR5 with 100 GB/s memory bandwidth (unified system memory, shared between CPU and GPU).
What is the Apple M2 (24GB) best for?
With 24 GB of VRAM, the Apple M2 (24GB) is well-suited for running 7B–32B models at Q4 with room for context, making it a great all-rounder for local LLM inference.
What LLMs can the Apple M2 (24GB) run locally?
The Apple M2 (24GB) can run 39 of the 84 open-weight models tracked by CanItRun natively in VRAM at 8k context. Top options include: Llama 3.1 8B Instruct at Q8_0, Llama 3.2 3B Instruct at FP32, Llama 3.2 1B Instruct at FP32.
Can the Apple M2 (24GB) run Llama 3.3 70B Instruct?
The Apple M2 (24GB) does not have enough VRAM to run Llama 3.3 70B Instruct. You would need more VRAM or a lower quantization level.
Show 14 more questions
Can the Apple M2 (24GB) run Qwen 3.6 27B?
Yes. The Apple M2 (24GB) runs Qwen 3.6 27B natively in VRAM at Q3_K_M quantization, achieving approximately 5.9 tokens per second.
Can the Apple M2 (24GB) run Llama 3.1 8B Instruct?
Yes. The Apple M2 (24GB) runs Llama 3.1 8B Instruct natively in VRAM at Q8_0 quantization, achieving approximately 8.4 tokens per second.
How much VRAM does the Apple M2 (16GB) have?
The Apple M2 (16GB) has 16 GB of LPDDR5 with 100 GB/s memory bandwidth (unified system memory, shared between CPU and GPU).
What is the Apple M2 (16GB) best for?
With 16 GB of VRAM, the Apple M2 (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 M2 (16GB) run locally?
The Apple M2 (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 M2 (16GB) run Llama 3.3 70B Instruct?
The Apple M2 (16GB) 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 M2 (16GB) run Qwen 3.6 27B?
The Apple M2 (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 M2 (16GB) run Llama 3.1 8B Instruct?
Yes. The Apple M2 (16GB) runs Llama 3.1 8B Instruct natively in VRAM at Q5_K_M quantization, achieving approximately 11.8 tokens per second.
How much VRAM does the Apple M2 (8GB) have?
The Apple M2 (8GB) has 8 GB of LPDDR5 with 100 GB/s memory bandwidth (unified system memory, shared between CPU and GPU).
What is the Apple M2 (8GB) best for?
With 8 GB of VRAM, the Apple M2 (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 M2 (8GB) run locally?
The Apple M2 (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 M2 (8GB) run Llama 3.3 70B Instruct?
The Apple M2 (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 M2 (8GB) run Qwen 3.6 27B?
The Apple M2 (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 M2 (8GB) run Llama 3.1 8B Instruct?
The Apple M2 (8GB) does not have enough VRAM to run Llama 3.1 8B Instruct. You would need more VRAM or a lower quantization level.