Can Mac Studio M5 Ultra 128 GB run DeepSeek-R1-0528-Qwen3-8B?

128 GB unified · 1228 GB/s memory bandwidth · DeepSeek-R1-0528-Qwen3-8B has 8.2B parameters.

Short answer

Yes, excellent for chat and acceptable or better for every workload we test.

Best quantization for chat: 6-bit (6.4 GB of weights) on LM Studio. Generation: 65–110 tok/s, leaving about 110 GB free for your other apps.

How it rates for each workload

Same model and machine, different jobs. Agents and long documents send far bigger prompts than chat, so they need much more speed and memory.

Workload	Verdict	Quant	Generation	First reply	Context
Chatwith Open WebUI	Excellent	6-bit	65–110 tok/s	under 1 s	8K	Details
Coding questionswith Continue	Comfortable	6-bit	60–100 tok/s	≈1 s	16K	Details
Coding in a repositorywith Aider	Acceptable	8-bit	40–70 tok/s	≈4 s	32K	Details
Agentic codingwith OpenCode	Acceptable	8-bit	40–65 tok/s	≈5 s	32K	Details
Long documents (64K)with Open WebUI	Comfortable	8-bit	30–45 tok/s	≈20 s	64K	Details

Speeds are bandwidth-based estimates; no direct benchmark exists for this pair. “First reply” is the time to read the workload’s opening prompt. How we calculate

Get it running

Copy-paste commands for the 8-bit build with a 64K context. Ollama and llama.cpp download the file straight from Hugging Face.

ollama run hf.co/unsloth/DeepSeek-R1-0528-Qwen3-8B-GGUF:Q8_0

Downloads 8.71 GB (Q8_0). Ollama's default context is small: set OLLAMA_CONTEXT_LENGTH=65536 before `ollama serve` for long prompts.

Other models for Mac Studio M5 Ultra 128 GB

The best-rated models for chat on this machine.

Everything Mac Studio M5 Ultra 128 GB can run

Cheapest machines for DeepSeek-R1-0528-Qwen3-8B

Hardware rated comfortable or better for chat, lowest price first.

Same chip, other memory: 256 GB, 512 GB

Different tool, context size or workload?

These pages use typical settings. The full check models your exact setup.

Open sourceCheck my exact setup