Mac Studio M3 Ultra 96GB
96 GB unified · 819 GB/s
Prices
Every retailer we track in Canada, cheapest first. Buy links may earn us a commission.
No retailer in this market lists it right now.
What it runs
Every model in the catalogue, with the best quant for this hardware. Estimates, not measurements.
| Model | Quant | Fits | Memory | Max context | Decode @32k | Prefill | Good for |
|---|---|---|---|---|---|---|---|
| Qwen 3.8 27B Alibaba Qwen | Q8_0 | fits | 30.0 GB | 256k | 17tok/sest. | 181 tok/s | chat |
| Qwen 3.5 122B-A10B Alibaba Qwen | Q3_K_M | fits | 54.2 GB | 256k | 58tok/sest. | 424 tok/s | chat, coding |
| Qwen 3.6 27B Alibaba Qwen | Q8_0 | fits | 29.6 GB | 256k | 17tok/sest. | 181 tok/s | chat |
| Qwen3 Coder Next Alibaba Qwen | Q6_K | fits | 62.7 GB | 256k | 78tok/sest. | 1021 tok/s | chat, coding, agents, long documents |
| gpt-oss 120B OpenAI | MXFP4 | fits | 61.0 GB | 128k | 74tok/sest. | 684 tok/s | chat, coding, agents, long documents |
| Nemotron 3 Super 120B-A12B NVIDIA | UD-Q3_K_M | fits | 58.7 GB | 256k | 50tok/sest. | 428 tok/s | chat, coding, long documents |
| Gemma 4 31B | Q8_0 | fits | 34.5 GB | 256k | 15tok/sest. | 161 tok/s | chat |
| Mistral Small 4 119B Mistral AI | UD-Q4_K_M | fits | 70.2 GB | 97k154k with q8 cache | 62tok/sest. | 471 tok/s | chat, coding |
| Qwen 3.6 35B-A3B Alibaba Qwen | Q8_0 | fits | 35.8 GB | 256k | 84tok/sest. | 1290 tok/s | chat, coding, agents, long documents |
| Qwen 3.5 27B Alibaba Qwen | Q8_0 | fits | 29.6 GB | 256k | 17tok/sest. | 181 tok/s | chat |
| Gemma 4 26B-A4B | Q8_0 | fits | 26.6 GB | 256k | 63tok/sest. | 1147 tok/s | chat, coding, agents, long documents |
| GLM 4.5 Air Z.ai | Q3_K_M | fits | 59.8 GB | 97k128k with q8 cache | 21tok/sest. | 158 tok/s | chat, coding |
| Qwen3 Next 80B-A3B Alibaba Qwen | Q6_K | fits | 62.6 GB | 256k | 97tok/sest. | 1215 tok/s | chat, coding, agents, long documents |
| Qwen 3.5 35B-A3B Alibaba Qwen | Q8_0 | fits | 35.8 GB | 256k | 84tok/sest. | 1290 tok/s | chat, coding, agents, long documents |
| GLM 4.7 Flash Z.ai | Q8_0 | fits | 32.1 GB | 198k | 45tok/sest. | 673 tok/s | chat, coding, agents, long documents |
| Devstral Small 2 24B Mistral AI | Q8_0 | fits | 29.1 GB | 296k384k with q8 cache | 18tok/sest. | 191 tok/s | chat |
| Nemotron 3.5 Lightning 30B-A3B NVIDIA | UD-Q6_K_XL | fits | 33.6 GB | 256k | 89tok/sest. | 1472 tok/s | chat, coding, agents, long documents |
| gpt-oss 20B OpenAI | MXFP4 | fits | 12.8 GB | 128k | 99tok/sest. | 963 tok/s | chat, coding, agents, long documents |
| Kimi Linear 48B-A3B Moonshot AI | Q8_0 | fits | 49.7 GB | 1024k | 91tok/sest. | 1112 tok/s | chat, coding, long documents |
| Qwen3 Coder 30B-A3B Alibaba Qwen | Q8_0 | fits | 34.0 GB | 256k | 50tok/sest. | 593 tok/s | chat, coding, agents |
| Qwen3 32B Alibaba Qwen | Q8_0 | fits | 41.2 GB | 40k | 12tok/sest. | 112 tok/s | – |
| Gemma 4 12B | Q8_0 | fits | 13.4 GB | 256k | 39tok/sest. | 419 tok/s | chat, coding, long documents |
| Qwen 3.5 9B Alibaba Qwen | Q8_0 | fits | 10.7 GB | 256k | 50tok/sest. | 522 tok/s | chat, coding, long documents |
| Mistral Small 3.2 24B Mistral AI | Q8_0 | fits | 29.1 GB | 128k | 18tok/sest. | 191 tok/s | chat |
| Gemma 3 27B | Q8_0 | fits | 30.4 GB | 128k | 17tok/sest. | 192 tok/s | chat |
| DeepSeek R1 Distill Qwen 32B DeepSeek | Q8_0 | fits | 41.2 GB | 128k | 12tok/sest. | 129 tok/s | – |
| Llama 3.3 70B Meta | Q5_K_M | fits | 57.3 GB | 78k128k with q8 cache | 9tok/sest. | 61 tok/s | – |
| Ministral 3 14B Mistral AI | Q8_0 | fits | 19.2 GB | 256k | 27tok/sest. | 290 tok/s | chat, coding |
| Qwen3 14B Alibaba Qwen | Q8_0 | fits | 20.4 GB | 40k | 25tok/sest. | 261 tok/s | chat, coding |
| DeepSeek R1 0528 Qwen3 8B DeepSeek | Q8_0 | fits | 13.4 GB | 128k | 39tok/sest. | 430 tok/s | chat, coding |
| Qwen3 8B Alibaba Qwen | Q8_0 | fits | 13.4 GB | 40k | 39tok/sest. | 430 tok/s | chat, coding |
| Phi-4 14B Microsoft | Q8_0 | fits | 21.5 GB | 16k | 24tok/sest. | 262 tok/s | chat, coding |
| Qwen 3.5 4B Alibaba Qwen | Q8_0 | fits | 6.0 GB | 256k | 95tok/sest. | 977 tok/s | chat, coding, long documents |
| Ministral 3 8B Mistral AI | Q8_0 | fits | 13.4 GB | 256k | 39tok/sest. | 415 tok/s | chat, coding |
| Granite 4.2 8B IBM | Q8_0 | fits | 14.5 GB | 128k | 36tok/sest. | 396 tok/s | chat, coding |
| Gemma 4 E4B | Q8_0 | fits | 9.3 GB | 128k | 58tok/sest. | 656 tok/s | chat, coding, long documents |
| LFM2.5 8B-A1B Liquid AI | Q8_0 | fits | 10.7 GB | 125k | 119tok/sest. | 2145 tok/s | chat, coding, long documents |
| Llama 3.1 8B Meta | Q6_K | fits | 10.9 GB | 128k | 49tok/sest. | 454 tok/s | chat, coding |
| DeepSeek V4 Flash DeepSeek | UD-Q2_K_XL | won't run | 93.7 GB | – | – | – | – |
| GLM 5.3 Flash Z.ai | UD-Q2_K_XL | won't run | 102.5 GB | – | – | – | – |
| Qwen 3.8 Flash Next Alibaba Qwen | UD-Q2_K_XL | won't run | 75.1 GB | – | – | – | – |
| MiniMax M2.7 MiniMax | UD-Q2_K_XL | won't run | 78.7 GB | 5k9k with q8 cache | – | – | – |
Specifications
From the manufacturer's published specifications; sources are linked where we have them.
- Chip
- Apple M3 Ultra (80-core GPU)
- Memory bus
- 1024-bit
- Memory bandwidth
- 819 GB/s
- FP16 compute
- 28 TFLOPS
- Software
- Metal
- Unified memory
- 96 GB
- GPU can use by default
- 72 GB
- GPU can use, tuned
- 88 GB
- Operating system
- macos
- Processor
- M3 Ultra 32-core
- Storage
- 1000 GB