# Apple Mac Studio M3 Ultra 512GB for local LLMs

> The original 512GB Mac for giant models, now available only used and at volatile prices.

The M3 Ultra 512GB has 512GB of unified memory (about 384GB usable by the GPU by default) and 819 GB/s of memory bandwidth, which decides how big a model fits and how fast it generates text. It comfortably runs models up to about 320B parameters: the largest in our catalog is GLM-5.3-Flash at Q6_K (273GB with 8K context, ~17 tokens/sec).

## Key specs

- Memory: 512GB LPDDR5 (unified; ~384GB usable by the GPU by default)
- Memory bandwidth: 819 GB/s
- Architecture: Apple M3 Ultra (2025)
- Power: 480W
- Typical price: $9,499 MSRP (checked 2026-10)
- Software: Metal, MLX

## Best models for this GPU

- **Best overall (also coding, reasoning, vision):** GLM-5.3-Flash at Q6_K, ~272.6GB with 8K context, ~17 tokens/sec
- **Runner-up:** DeepSeek V4 Flash at FP4, ~167.5GB with 8K context, ~32 tokens/sec
- **Fastest good model:** gpt-oss-20b at MXFP4, ~15.0GB with 8K context, ~80 tokens/sec
- **Largest model that fits:** GLM-5.3 at Q3_K_M, ~374.9GB with 8K context, ~13 tokens/sec

## Compatibility (best quantization that fits, 8K context)

| Model | Params | Quant | Memory | Speed | Verdict |
| --- | --- | --- | --- | --- | --- |
| Gemma 4 E4B | 8B | Q8_0 | 10.5GB | ~44 tok/s | Runs well |
| Qwen3.5 9B | 9B | Q8_0 | 12.0GB | ~39 tok/s | Runs well |
| Phi-4 | 14.7B | Q8_0 | 18.3GB | ~28 tok/s | Runs well |
| gpt-oss-20b | 20.9B (3.6B active) | MXFP4 | 15.0GB | ~80 tok/s | Runs well |
| Gemma 4 26B A4B | 26B (3.8B active) | Q8_0 | 29.5GB | ~54 tok/s | Runs well |
| Qwen3.8 27B | 27B | Q8_0 | 31.6GB | ~15 tok/s | Runs well |
| Gemma 4 31B | 31B | Q8_0 | 35.9GB | ~14 tok/s | Runs well |
| Qwen3.6 35B-A3B | 35B (3B active) | Q8_0 | 41.0GB | ~62 tok/s | Runs well |
| GLM-4.5-Air | 106B (12B active) | Q8_0 | 118.1GB | ~20 tok/s | Runs well |
| Llama 4 Scout | 109B (17B active) | Q8_0 | 120.2GB | ~14 tok/s | Runs well |
| gpt-oss-120b | 117B (5.1B active) | MXFP4 | 68.3GB | ~70 tok/s | Runs well |
| Qwen3.5 122B-A10B | 122B (10B active) | Q8_0 | 134.3GB | ~23 tok/s | Runs well |
| DeepSeek V4 Flash | 284B (13B active) | FP4 | 167.5GB | ~32 tok/s | Runs well |
| GLM-5.3-Flash | 320B (18B active) | Q6_K | 272.6GB | ~17 tok/s | Runs well |
| GLM-5.3 | 744B (40B active) | Q3_K_M | 374.9GB | ~13 tok/s | Fits, tight |

## Pros
- 512GB of unified memory
- 819 GB/s bandwidth
- Quiet and efficient for its capacity

## Cons
- Discontinued, with volatile used prices
- Slow prompt processing on long inputs
- No CUDA support

Full page: https://llamabuilds.ai/gpus/mac-studio-m3-ultra-512gb · Best models: https://llamabuilds.ai/gpus/mac-studio-m3-ultra-512gb/best-models

Sources: [Apple Mac Studio (2025) tech specs](https://support.apple.com/en-us/122211), [MacRumors: Apple cuts more Mac Studio and Mac mini RAM options](https://www.macrumors.com/2026/05/05/apple-mac-studio-mac-mini-ram-cuts/), [Apple Mac Studio power consumption and thermal output](https://support.apple.com/en-us/102027)

---
Source: https://llamabuilds.ai/gpus/mac-studio-m3-ultra-512gb · Updated October 14, 2026 · Llama Builds. Speeds marked ~ are estimates from our bandwidth model (https://llamabuilds.ai/methodology); measured results are labeled as such.