CanItRun Logocanitrun.

Apple M5

The Apple M5 ships in 16–32 GB unified-memory configurations at 153 GB/s. Across those configurations it runs 46 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:32 GB16 GB
ConfigurationBandwidthCPU coresNative models+ Offload
32 GB153 GB/s10 (4S + 6E)46 / 840
16 GB153 GB/s10 (4S + 6E)26 / 840
Vendor: Apple
Memory type: LPDDR5X
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 M5'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.

Base-tier bandwidth, from M5's own vantage point

Apple's cheapest M-series chip doesn't get a bandwidth bump every year — M3 shipped at the exact same 100 GB/s as M2, a full generation with no change at all. Tracking the base tier's memory bandwidth across five generations:

Apple M1 (16GB)
68 GB/s
Apple M2 (16GB)
100 GB/s
Apple M3 (16GB)
100 GB/s
Apple M4 (16GB)
120 GB/s
Apple M5 (16GB) — this page
153 GB/s

M1's 68 GB/s jumped 47% to M2's 100 GB/s, M3 held flat at that same 100 GB/s, and M4 resumed the climb with a 20% jump to 120 GB/s. M5 (this page) adds the largest single-generation increase in the base tier's five-generation history, up 27.5% to 153 GB/s. Since decode is bandwidth-bound, that compounds directly into tokens/sec: this site's calculator measures Llama 3.1 8B at 18.1 tok/s on this chip versus 14.2 tok/s on the identically-sized M4 16GB — a real generational leap, not a spec-sheet rounding error.

Apple M5 (32GB)

With 32 GB LPDDR5X at 153 GB/s, this configuration runs 46 models natively. It comfortably runs 7B–32B models at Q4; 70B-class models typically need CPU offload.

Apple M5 (32GB): Apple's entry-level M5 chip at its maximum tracked memory configuration, available in the MacBook Pro since October 2025 and the MacBook Air since March 2026, with a 10-core CPU (4 super cores + 6 efficiency cores) and 153 GB/s of unified-memory bandwidth — a 27.5% jump over the M4 base chip's 120 GB/s, the largest single-generation bandwidth increase anywhere in the base tier's five-generation history (see this page's bandwidth note).

With 24 GB of real headroom, this configuration comfortably clears genuine 30B-class models: Qwen3 32B fits at Q4_K_M (23.88 GB, 5.7 tok/s) — about 27% faster than the identically-sized M4 32GB on the same fit (4.5 tok/s there), tracking the 153-vs-120 GB/s bandwidth gap closely — and Qwen 3.6 35B reaches Q3_K_M (21.26 GB, 6.4 tok/s). The MoE-vs-dense gap is stark at this capacity too: Qwen 3.5 35B-A3B — a bigger model by parameter count than Qwen3 32B — fits at Q3_K_M (19.04 GB) at 24.6 tok/s, more than 4x Qwen3 32B's speed, since MoE decode only reads its active parameters off the memory bus. 46 of the 84 tracked models fit natively.

MLX and llama.cpp's Metal backend are both mature on this chip. This configuration's largest fits (~19-24 GB) stay under the default ~24 GB Metal working-set limit (75% of 32 GB), so the manual wired_limit override rarely matters here.

Models the 32 GB configuration runs natively (46)

Show 41 more

Apple M5 (16GB)

With 16 GB LPDDR5X at 153 GB/s, this configuration runs 26 models natively. It handles smaller models (7B–14B) at Q4–Q5 quantization.

Apple M5 (16GB): the base configuration of Apple's entry-level M5 chip, launched October 2025 as the standard memory on the MacBook Pro, and since March 2026 the MacBook Air too, with the same 10-core CPU (4 super cores + 6 efficiency cores) and 153 GB/s bandwidth as the 32GB build — only the memory pool changes.

With 8 GB of real headroom after the standard 8 GB reservation, this site's calculator fits Llama 3.1 8B at Q5_K_M (7.58 GB, 18.1 tok/s) and Qwen2.5 7B at Q6_K (7.51 GB, 18.2 tok/s) — about 27.5% faster than the identically-sized M4 16GB on the same two models (14.2 and 14.3 tok/s there), tracking the 153-vs-120 GB/s bandwidth ratio almost exactly. 26 of the 84 tracked models fit natively — the same count as the base M1 through M4 16GB configurations at this identical 8 GB headroom, since capacity, not bandwidth, decides the model list.

MLX and llama.cpp's Metal backend are both mature here. 16GB remains the practical minimum for serious local LLM inference on Apple Silicon — worth confirming before buying, since it's still the standard configuration on the entry MacBook Air and MacBook Pro.

Models the 16 GB configuration runs natively (26)

Show 21 more

Too large for any Apple M5 configuration (38)

Continue reading

hardware11 min
Apple Silicon for LLMs: M1-M5 Complete Guide

Frequently asked questions

How much memory does the Apple M5 have?
The Apple M5 ships in 2 unified-memory configurations: 32 GB and 16 GB, all at 153 GB/s.
Should I get the 16 GB or 32 GB Apple M5?
Both run everything that fits natively in 16 GB. The extra memory in the 32 GB configuration additionally fits Mixtral 8x7B Instruct v0.1, Qwen 3.5 35B-A3B (MoE), Qwen 3.6 35B, and 17 more models natively in VRAM — worth the upgrade if you plan to run any of those.
How much VRAM does the Apple M5 (32GB) have?
The Apple M5 (32GB) has 32 GB of LPDDR5X with 153 GB/s memory bandwidth (unified system memory, shared between CPU and GPU).
What is the Apple M5 (32GB) best for?
With 32 GB of VRAM, the Apple M5 (32GB) 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 M5 (32GB) run locally?
The Apple M5 (32GB) can run 46 of the 84 open-weight models tracked by CanItRun natively in VRAM at 8k context. Top options include: Llama 3.1 8B Instruct at BF16, Llama 3.2 3B Instruct at FP32, Llama 3.2 1B Instruct at FP32.
Can the Apple M5 (32GB) run Llama 3.3 70B Instruct?
The Apple M5 (32GB) 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 M5 (32GB) run Qwen 3.6 27B?
Yes. The Apple M5 (32GB) runs Qwen 3.6 27B natively in VRAM at Q5_K_M quantization, achieving approximately 6.2 tokens per second.
Can the Apple M5 (32GB) run Llama 3.1 8B Instruct?
Yes. The Apple M5 (32GB) runs Llama 3.1 8B Instruct natively in VRAM at BF16 quantization, achieving approximately 7.2 tokens per second.
How much VRAM does the Apple M5 (16GB) have?
The Apple M5 (16GB) has 16 GB of LPDDR5X with 153 GB/s memory bandwidth (unified system memory, shared between CPU and GPU).
What is the Apple M5 (16GB) best for?
With 16 GB of VRAM, the Apple M5 (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 M5 (16GB) run locally?
The Apple M5 (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 M5 (16GB) run Llama 3.3 70B Instruct?
The Apple M5 (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 M5 (16GB) run Qwen 3.6 27B?
The Apple M5 (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 M5 (16GB) run Llama 3.1 8B Instruct?
Yes. The Apple M5 (16GB) runs Llama 3.1 8B Instruct natively in VRAM at Q5_K_M quantization, achieving approximately 18.1 tokens per second.