Is the M3 faster than the M2 for LLMs?

About the same: their memory bandwidth is within a few percent. Both go up to 24 GB of memory.

Which can run bigger models, the M2 or the M3?

Neither: both go up to 24 GB. At that size the M2 runs 20 and the M3 20 of 73 open models well; the faster one clears more of them at 10 tokens a second.

M2 vs M3 for local LLMs: memory, speed, which to buy

Side by side

	M2	M3
Memory sizes	8, 16, 24 GB	8, 16, 24 GB
Memory bandwidth	100 GB/s	100 GB/s
For models, at most	about 18 GB	about 18 GB
Models that run well	20 of 73	20 of 73
MLX seat class	Core	Core
Found in	MacBook Air (M2, 2022), MacBook Pro (13-inch, M2, 2022), MacBook Air (15-inch, M2, 2023), Mac mini (2023)	MacBook Pro (14-inch, M3, Nov 2023), iMac (24-inch, M3, 2023, Two ports), iMac (24-inch, M3, 2023, Four ports), MacBook Air (13-inch, M3, 2024), MacBook Air (15-inch, M3, 2024)

Speed with popular models

Tokens a second for one conversation, each chip at its largest memory size (24 GB and 24 GB), with the best version of the model that fits.

Model	M2	M3
Llama 3.1 8B Instruct8B	10–18 4-bit	10–18 4-bit
gpt-oss-20b21B · 3.6B active	24–42 native 4.25-bit	24–42 native 4.25-bit
Qwen3.6 35B A3B35B · 3B active	35–62 3-bit lower quality	35–62 3-bit lower quality
Qwen3.8 27B27B	3–5 4-bit	3–5 4-bit
Llama 3.3 70B Instruct70B	too big	too big
gpt-oss-120b117B · 5.1B active	too big	too big
Qwen3.5 122B A10B122B · 10B active	too big	too big
DeepSeek V4 Flash 0731284B · 13B active	too big	too big

Questions

Is the M3 faster than the M2 for LLMs?: About the same: their memory bandwidth is within a few percent. Both go up to 24 GB of memory.
Which can run bigger models, the M2 or the M3?: Neither: both go up to 24 GB. At that size the M2 runs 20 and the M3 20 of 73 open models well; the faster one clears more of them at 10 tokens a second.

More comparisons

The M2 for local AI · The M3 for local AI · Compare every model · The best Mac for local LLMs

How it’s worked out

Memory: the model’s parameters at 8, 6, 4, 3 or 2 bits per weight plus MLX’s group scales (half a bit per weight), or its native format, with room for the context on top. By default macOS lets the GPU use about 75% of unified memory. Speed: memory bandwidth divided by the bytes read per token is the ceiling for one conversation; in llama.cpp’s public Apple Silicon benchmarks real runs reach about 45% to 80% of it, which is the range shown. These are estimates, not measurements: measure your own Mac on MLX Bench. Model figures from each official model card, chip figures from Apple’s Tech Specs. Checked 2026-10-19. Same data as mlx.app/compare.