What is the best local LLM for a Mac Studio M3 Ultra?

With its huge unified memory, the M3 Ultra runs Llama 3.3 70B at Q8 (near-full quality) and 100B+ MoE models like gpt-oss 120B locally at ~25-30 tokens/sec. For OpenClaw you can even keep several models resident at once for multi-agent setups.

How much unified memory does the M3 Ultra have?

It depends on when you bought it. The M3 Ultra Mac Studio originally configured up to 512 GB of unified memory at roughly 800 GB/s. Apple removed the 512 GB option in early March 2026 and the 256 GB option on May 5, 2026, and it discontinued the M3 Ultra on August 25, 2026. Every M3 Ultra is now used. 96 GB still holds a 70B model at Q8 and a 120B-class MoE at Q4. For 256 GB or 512 GB new, the M5 Ultra Mac Studio configures to both.

Is the M3 Ultra fast for local LLMs?

For its class, yes. The ~800 GB/s bandwidth delivers ~25-30 tokens/sec on 70B-class models — much faster than an M4 Pro and enough for comfortable interactive use, though still below datacenter GPUs on raw throughput.

M3 Ultra vs M4 Max for local AI?

The M3 Ultra wins on bandwidth (~800 vs ~410 GB/s), so it runs big models faster. Apple discontinued both on August 25, 2026, so both are used purchases now. The M4 Max is much cheaper and plenty for 70B at Q4; choose the Ultra when you need 70B at Q8, a 120B-class MoE, or multiple resident models.

← All guides

Hardware July 7, 2026

Best Local LLM for Mac Studio M3 Ultra (2026): Used Only, 96GB to 512GB

The Mac Studio M3 Ultra runs unified memory at ~800 GB/s and puts models on your desk that no single consumer GPU can touch. One important update since this page was written: Apple removed the 512GB configuration in March 2026 and the 256GB configuration on May 5, 2026. Apple then discontinued the M3 Ultra on August 25, 2026, when the M5 Ultra Mac Studio replaced it. Every M3 Ultra is now a used or clearance unit.

Picking hardware for an OpenClaw host?

Use the local model calculator first, then see our AI training options if you want help matching your workload to the right rig.

Short answer: the M3 Ultra is still the top consumer local-LLM machine for sheer model size, and it is now a used-market machine. It runs Llama 3.3 70B at Q8, 100B+ MoE models (gpt-oss 120B, Mistral Small MoE), and very long context — all locally at ~25-30 tok/sec. The ~800 GB/s bandwidth makes it noticeably faster than an M4 Max for big models.

Availability update, August 2026. Apple pulled the 512 GB M3 Ultra configuration in early March 2026 and the 256 GB configuration on May 5, 2026, in the middle of the global DRAM shortage. Apple then discontinued the M3 Ultra on August 25, 2026. The new Mac Studio uses the M5 Max or the M5 Ultra, and the M5 Ultra configures to 96 GB, 256 GB or 512 GB. Every M3 Ultra is now used or refurbished, and big-memory units price on scarcity rather than depreciation. Everything below still applies to the M3 Ultra you can buy used. Details: best models for the biggest Mac Studio.

The Memory Math

96 GB of unified memory is enough for 70B at Q8, a 120B-class MoE at Q4, and one large model plus a small router model resident at the same time. That covers most single-operator OpenClaw setups. What 96 GB does not do is the thing the 512 GB machine was famous for — holding a 400B-class model at native precision, or keeping four large models loaded for a multi-agent fleet. If that is your requirement, buy a used 512 GB M3 Ultra, configure a new M5 Ultra to 512 GB, or look at a DGX Spark cluster.

Ready to buy? See the tested hardware list with current prices.

What Actually Fits (Model Picks)

Model	Quant	Memory	Speed	Notes
Llama 3.3 70B	Q8_0	~75 GB	~25 tok/s	70B at near-full quality
gpt-oss 120B (MoE)	Q4/Q5	~70-90 GB	~28 tok/s	Flagship MoE, runs locally
Qwen 3.6 27B	Q8_0 + huge ctx	~30 GB	~40 tok/s	27B with massive context
Multiple models resident	mixed	fits easily	—	Run several at once for agents

What You Can’t Run

Beating a datacenter GPU on raw throughput — Apple bandwidth is high for a desktop but below an H100-class card; expect ~25-30 tok/s, not hundreds.
Training large models — this is an inference machine, not a training rig.
Cheap — the value case is “runs models a $30k GPU box would,” not low price.

🎮 THE BIG-MEMORY APPLE SILICON PICK

Apple discontinued the M3 Ultra Mac Studio on August 25, 2026, so it is a used-market machine now — still a fast big-memory machine you can put on a desk, though used 256 GB and 512 GB units are scarce. A 48 GB+ Mac is the entry to 70B; the Mac mini M4 is the budget always-on host for smaller models, starting at $799 for 16 GB now that Apple has dropped the 256 GB storage SKU.

48GB+Premium Mac 48 GB+ ↗ MINIMac mini M4 ↗

OpenClaw Setup

Point OpenClaw at your local model through Ollama:

# pull and run your pick, then set it as the OpenClaw default
ollama pull llama3.3:70b
openclaw config set agents.defaults.models.chat "ollama/llama3.3:70b"

For agent reliability, prefer a model with clean tool-call output (gpt-oss 20B where it fits) and cap context to what your memory holds. See the tool-calling reliability guide.

Before you order parts, check the tested hardware list for current prices by tier.