The best local LLM for a Mac Studio M4 Ultra (96 GB) is Qwen 3.6-27B at 73 tok/s. With 96 GB of unified memory it runs 55 of the models we benchmark — from compact options up to 276B-class models. For everyday chat and coding, Qwen 3.6-27B is the sweet spot. Full ranking below.
Ranked by LLMCheck suitability (capability balanced against speed on the M4 Ultra). 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 | Qwen 3.6-27B | 27B | Apache 2.0 | 73 tok/s est. | 44/50 |
| 2 | Inkling-Small | 276B | Apache 2.0 | 40 tok/s est. | 46/50 |
| 3 | KAT-Coder-V2.5 | 35B | Apache 2.0 | 95 tok/s est. | 39/50 |
| 4 | Qwen 3.6-35B-A3B | 35B | Apache 2.0 | 95 tok/s est. | 38/50 |
| 5 | Nemotron 3.5 Lightning | 30B | OpenMDW | 109 tok/s est. | 36/50 |
| 6 | Laguna S 2.1 | 118B | OpenMDW | 47 tok/s est. | 43/50 |
| 7 | Muse Glimmer 30B | 30B | Apache 2.0 | 49 tok/s est. | 42/50 |
| 8 | Gemma 4 26B-A4B | 26B | Apache 2.0 | 87 tok/s est. | 35/50 |
| 9 | Gemma 4 31B | 31B | Apache 2.0 | 44 tok/s est. | 40/50 |
| 10 | Laguna XS 2.1 | 33B | OpenMDW | 100 tok/s est. | 33/50 |
| 11 | Nemotron-Cascade 2 | 30B | NVIDIA Open | 106 tok/s est. | 30/50 |
| 12 | GLM-4.7-Flash | 31B | MIT | 64 tok/s est. | 35/50 |
Showing the top 12 of 55 models that fit in 96 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 Qwen 3.6-27B on M4 Ultra benchmark page for exact settings.
The Mac Studio M4 Ultra (96 GB) comfortably runs 55 of the models we benchmark, led by Qwen 3.6-27B. Grab one and start running LLMs offline today:
As an Amazon Associate, LLMCheck earns from qualifying purchases. Affiliate links cost you nothing extra and never influence our rankings.
Qwen 3.6-27B (27B, Apache 2.0) is the best all-round pick at 73 tok/s on the M4 Ultra. If you want maximum speed, Maple Preview 20B-A1B hits 510 tok/s; for maximum capability, Inkling-Small still fits in 96 GB.
About 55 of the 80 models in the LLMCheck leaderboard fit in 96 GB of unified memory, from compact models up to Inkling-Small (276B).
Yes. A 70B model in Q4 quantization needs roughly 40–44 GB of memory, which fits in 96 GB with headroom for context.
96 GB is plenty for local AI — you can run capable 30B–70B-class models. Because Apple Silicon uses unified memory, that figure is both your system RAM and your VRAM.