The best local LLM for a MacBook Air M2 (16 GB) is Bonsai 27B at 12 tok/s. With 16 GB of unified memory it runs 25 of the models we benchmark — from compact options up to 27B-class models. For everyday chat and coding, Bonsai 27B is the sweet spot. Full ranking below.
Ranked by LLMCheck suitability (capability balanced against speed on the M2). Click a model for its full benchmark and setup. All speeds are index estimates (memory-bandwidth model, cross-referenced with sourced benchmarks where available) — submit a real run →
| # | Model | Size | License | Speed | Capability |
|---|---|---|---|---|---|
| 1 | Bonsai 27B | 27B | Apache 2.0 | 12 tok/s | 33/50 |
| 2 | Nanbeige4.2-3B | 3B | Apache 2.0 | 68 tok/s | 20/50 |
| 3 | Mistral Small 3.2 24B | 24B | Apache 2.0 | 7 tok/s est. | 25/50 |
| 4 | Maple Preview 20B-A1B | 20B | MIT | 47 tok/s est. | 20/50 |
| 5 | Devstral Small 24B | 24B | Apache 2.0 | 6 tok/s est. | 24/50 |
| 6 | DeepSeek R1 8B | 8B | MIT | 58 tok/s | 16/50 |
| 7 | Phi-4 Mini | 3.8B | MIT | 72 tok/s | 14/50 |
| 8 | Phi-4 14B | 14B | MIT | 28 tok/s | 19/50 |
| 9 | Qwen 3 8B | 8B | Apache 2.0 | 55 tok/s | 15/50 |
| 10 | Qwen 3 14B | 14B | Apache 2.0 | 9 tok/s est. | 20/50 |
| 11 | Gemma 3 12B | 12B | Gemma | 32 tok/s | 17/50 |
| 12 | Qwen 3.5 9B | 9B | Apache 2.0 | 17 tok/s est. | 18/50 |
Showing the top 12 of 25 models that fit in 16 GB. See the full leaderboard or all benchmarks.
The fastest way to get started is Ollama. Install it, then pull the top pick for your Mac:
Prefer a GUI? LM Studio gives you a one-click download and chat window. For step-by-step help see our Ollama install guide, or open the Bonsai 27B on M2 benchmark page for exact settings.
The MacBook Air M2 (16 GB) tops out at Bonsai 27B. Newer Apple Silicon with more unified memory runs larger, smarter models much faster:
As an Amazon Associate, LLMCheck earns from qualifying purchases. Affiliate links cost you nothing extra and never influence our rankings.
Bonsai 27B (27B, Apache 2.0) is the best all-round pick at 12 tok/s on the M2. If you want maximum speed, SmolLM3 3B hits 78 tok/s; for maximum capability, Mistral Small 3.2 24B still fits in 16 GB.
About 25 of the 80 models in the LLMCheck leaderboard fit in 16 GB of unified memory, from compact models up to Bonsai 27B (27B).
Not comfortably. A 70B model in Q4 needs ~40–44 GB; with 16 GB you should stick to models up to ~10 GB, such as Bonsai 27B. For 70B, look at a 48 GB+ Mac.
16 GB is great for small-to-mid models (up to ~14B comfortably); for 30B+ you'll want 32 GB or more. Because Apple Silicon uses unified memory, that figure is both your system RAM and your VRAM.