Is the M3 Pro faster than the M2 Pro for LLMs?

The M2 Pro reads memory about 33% faster than the M3 Pro (200 vs 150 GB/s), so it writes about 33% faster with any model that fits both. Reading a long prompt depends on GPU compute instead, which this comparison doesn’t cover.

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

The M3 Pro: The M3 Pro goes up to 36 GB of memory, the M2 Pro to 32 GB. At its largest it runs 28 of 73 open models well, against 30.

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

Side by side

	M2 Pro	M3 Pro
Memory sizes	16, 32 GB	18, 36 GB
Memory bandwidth	200 GB/s	150 GB/s
For models, at most	about 24 GB	about 27 GB
Models that run well	30 of 73	28 of 73
MLX seat class	Pro	Pro
Found in	MacBook Pro (14-inch, 2023), MacBook Pro (16-inch, 2023), Mac mini (2023)	MacBook Pro (14-inch, M3 Pro or M3 Max, Nov 2023), MacBook Pro (16-inch, Nov 2023)

Speed with popular models

Tokens a second for one conversation, each chip at its largest memory size (32 GB and 36 GB), with the best version of the model that fits.

Model	M2 Pro	M3 Pro
Llama 3.1 8B Instruct8B	11–19 8-bit	10–18 6-bit
gpt-oss-20b21B · 3.6B active	47–84 native 4.25-bit	36–63 native 4.25-bit
Qwen3.6 35B A3B35B · 3B active	53–94 4-bit	40–71 4-bit
Qwen3.8 27B27B	4–7 6-bit	3–5 6-bit
Llama 3.3 70B Instruct70B	4–7 2-bit lower quality	3–5 2-bit lower quality
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 Pro faster than the M2 Pro for LLMs?: The M2 Pro reads memory about 33% faster than the M3 Pro (200 vs 150 GB/s), so it writes about 33% faster with any model that fits both. Reading a long prompt depends on GPU compute instead, which this comparison doesn’t cover.
Which can run bigger models, the M2 Pro or the M3 Pro?: The M3 Pro: The M3 Pro goes up to 36 GB of memory, the M2 Pro to 32 GB. At its largest it runs 28 of 73 open models well, against 30.

More comparisons

The M2 Pro for local AI · The M3 Pro 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.