CanItRun Logocanitrun.

Apple M6

The Apple M6 ships in 16–32 GB unified-memory configurations at 170 GB/s. Across those configurations it runs 52 of our 94 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 GB170 GB/s12 (2S + 4P + 6E)52 / 940
16 GB170 GB/s12 (2S + 4P + 6E)27 / 940
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 M6'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 M6 (32GB)

With 32 GB LPDDR5X at 170 GB/s, this configuration runs 52 models natively. It comfortably runs 7B–32B models at Q4; 70B-class models typically need CPU offload.

Apple M6 (32GB) is a mobile/laptop Apple Silicon chip based on the Apple M6 architecture. Released in 2026. It features 32 GB of LPDDR5X unified memory at 170 GB/s memory bandwidth. As an Apple Silicon chip, its memory is unified between CPU and GPU, so the full 32 GB can be allocated to model weights. MLX gives the best performance on Apple Silicon; llama.cpp Metal backend is a solid alternative. Both are well-supported by Ollama.

For local LLM inference, this GPU runs 52 of the 94 models we track natively in VRAM at 8K context. The largest model it handles in VRAM is Mixtral 8x7B Instruct v0.1 (7.8 t/s at Q2_K). It comfortably runs models up to ~27-32B parameters at Q4. Larger models need CPU offload or multi-GPU. On Qwen 3.6 27B, it achieves approximately 6.9 tokens per second at Q5_K_M quantization.

Apple's Metal backend is fully supported by MLX and llama.cpp, giving excellent performance on macOS. Among laptop GPUs, it sits above Apple M5 (32GB) and Apple M4 (32GB) in performance, but below Apple M2 Pro (32GB).

Models the 32 GB configuration runs natively (52)

Show 47 more

Apple M6 (16GB)

With 16 GB LPDDR5X at 170 GB/s, this configuration runs 27 models natively. It handles smaller models (7B–14B) at Q4–Q5 quantization.

Apple M6 (16GB) is a mobile/laptop Apple Silicon chip based on the Apple M6 architecture. Released in 2026. It features 16 GB of LPDDR5X unified memory at 170 GB/s memory bandwidth. As an Apple Silicon chip, its memory is unified between CPU and GPU, so the full 16 GB can be allocated to model weights. MLX gives the best performance on Apple Silicon; llama.cpp Metal backend is a solid alternative. Both are well-supported by Ollama.

For local LLM inference, this GPU runs 27 of the 94 models we track natively in VRAM at 8K context. The largest model it handles in VRAM is Bonsai 27B (25.1 t/s at 1-bit (Q1_0)). It handles smaller models up to ~7-14B at reasonable precision, with some 27-32B models fitting at lower quantization. On Qwen3 8B, it achieves approximately 19.7 tokens per second at Q5_K_M quantization.

Apple's Metal backend is fully supported by MLX and llama.cpp, giving excellent performance on macOS. Among laptop GPUs, it sits above Apple M5 (16GB) and Apple M3 Pro (18GB) in performance, but below Apple M2 Pro (16GB).

Models the 16 GB configuration runs natively (27)

Show 22 more

Too large for any Apple M6 configuration (42)

Continue reading

hardware14 min
Apple Silicon for LLMs: M1 to M6 Complete Guide (2026)

Frequently asked questions

How much memory does the Apple M6 have?
The Apple M6 ships in 2 unified-memory configurations: 32 GB and 16 GB, all at 170 GB/s.
Should I get the 16 GB or 32 GB Apple M6?
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 22 more models natively in VRAM, worth the upgrade if you plan to run any of those.
How much VRAM does the Apple M6 (32GB) have?
The Apple M6 (32GB) has 32 GB of LPDDR5X with 170 GB/s memory bandwidth (unified system memory, shared between CPU and GPU).
What is the Apple M6 (32GB) best for?
With 32 GB of VRAM, the Apple M6 (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 M6 (32GB) run locally?
The Apple M6 (32GB) can run 52 of the 94 open-weight models tracked by CanItRun natively in VRAM at 8k context. Top options include: Qwen 3.8 27B at Q5_K_M, Muse Glimmer 30B at Q5_K_M, Ornith 1.5 9B at BF16.
Can the Apple M6 (32GB) run Gemma 4 31B?
Yes. The Apple M6 (32GB) runs Gemma 4 31B natively in VRAM at Q4_K_M quantization, achieving approximately 6.7 tokens per second.
Show 8 more questions
Can the Apple M6 (32GB) run Qwen 3.6 27B?
Yes. The Apple M6 (32GB) runs Qwen 3.6 27B natively in VRAM at Q5_K_M quantization, achieving approximately 6.9 tokens per second.
Can the Apple M6 (32GB) run Qwen3 8B?
Yes. The Apple M6 (32GB) runs Qwen3 8B natively in VRAM at BF16 quantization, achieving approximately 7.9 tokens per second.
How much VRAM does the Apple M6 (16GB) have?
The Apple M6 (16GB) has 16 GB of LPDDR5X with 170 GB/s memory bandwidth (unified system memory, shared between CPU and GPU).
What is the Apple M6 (16GB) best for?
With 16 GB of VRAM, the Apple M6 (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 M6 (16GB) run locally?
The Apple M6 (16GB) can run 27 of the 94 open-weight models tracked by CanItRun natively in VRAM at 8k context. Top options include: Ornith 1.5 9B at Q5_K_M, Qwen 3.5 9B at Q5_K_M, Bonsai 27B at 1-bit (Q1_0).
Can the Apple M6 (16GB) run Gemma 4 31B?
The Apple M6 (16GB) does not have enough VRAM to run Gemma 4 31B. You would need more VRAM or a lower quantization level.
Can the Apple M6 (16GB) run Qwen 3.6 27B?
The Apple M6 (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 M6 (16GB) run Qwen3 8B?
Yes. The Apple M6 (16GB) runs Qwen3 8B natively in VRAM at Q5_K_M quantization, achieving approximately 19.7 tokens per second.