1792 GB/s (+992)
B
Raises estimated decode speed by about 226%.40.1 tok/s decode
Raises estimated decode speed by about 226%.
~$9,999 MSRP
Can DeepSeek LLM 67B run on Mac Studio M2 Ultra 128GB?
YES — Runs Great
B60Good
○Estimated from fit model
DeepSeek LLM 67B needs ~61.4 GB VRAM. Mac Studio M2 Ultra 128GB has 92.2 GB. With Q4_K_M quantization, expect ~12 tok/s.
Runtime: llama.cppCapacity: RoomyBandwidth: HighStack: StandardBottleneck: Balanced
Share:
Operating mode
Choose the run profile you care about
Interactive favors responsiveness, while light API and scale-out lean harder on serving readiness. The fit stays the same, but the recommendation lens changes.
Current mode
Balanced
Balanced for general local use. Keeps the ranking neutral across personal and serving workflows.
Select quantization to explore
Q4_K_M (Medium quality) — 61.4 GB, 12.3 tok/s, Runs well
61.4 GB required92.2 GB available
67% VRAM used
Fit status
Runs well
Decode
12.3 tok/s
TTFT
15681 ms
Safe context
4K
Memory
61.4 GB / 92.2 GB
Memory breakdown
Weights40.9 GB
KV Cache5.8 GB
Runtime0.9 GB
Headroom13.8 GB
See how fast it feels
See how fast it feelsDeepSeek LLM 67B on Mac Studio M2 Ultra 128GB
1st promptCold start — includes initialization
>Solve: If a train travels at 60 mph for 2.5 hours, how far does it go?
Let me work through this step by step.
Given information:
- Speed = 60 mph (miles per hour)
- Time = 2.5 hours
Formula: Distance = Speed x Time
Calculation:
Distance = 60 mph x 2.5 h
= 150 miles
The train travels 150 miles.
Double-check: at 60 mph the train covers 60 miles each hour. In half an hour it covers 30 miles. So in 2.5 hours: 60 + 60 + 30 = 150 miles. Confirmed.
2nd promptWarm — model ready, no init overhead
>What if the train stops for 20 minutes halfway?
The stop doesn't change the total distance covered while moving, but it affects total trip time.
Breakdown:
- First half: 1.25 hours at 60 mph = 75 miles
- Stop: 20 minutes (no distance)
- Second half: 1.25 hours at 60 mph = 75 miles
Total distance: 150 miles (same)
Total time: 2.5 h + 0.33 h = 2.83 hours
Average speed: 150 / 2.83 = 53 mph
The distance stays the same but average speed drops to 53 mph because of the stop.
Estimated: 12.3 tok/s decode · 15.7s TTFT (warm) · 31 tok/s prefill
What limits this setup
This setup is broadly balanced for this model.
Shared-memory contention still exists
The OS, browser, and inference runtime all compete for the same physical memory pool, so real-world headroom is less forgiving than raw capacity suggests.
Best improvement path
Performance by workload
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Chat | B | Runs well | 12.3 tok/s | 8553 ms | 4K |
| Coding | B | Runs well | 12.3 tok/s | 15681 ms | 4K |
| Agentic Coding | B | Runs well | 12.3 tok/s | 22808 ms | 4K |
| Reasoning | B | Runs well | 12.3 tok/s | 18532 ms | 4K |
| RAG | B | Runs well | 12.3 tok/s | 28510 ms | 4K |
Quantization options
How DeepSeek LLM 67B (67B params) fits at each quantization level on Mac Studio M2 Ultra 128GB (92.2 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 26.1 GB | Low | C52 |
Q3_K_S | 3 | 32.8 GB | Low | C54 |
NVFP4 | 4 | 37.5 GB | Medium | C55 |
Q4_K_M | 4 | 40.9 GB | Medium | B55 |
Q5_K_M | 5 | 48.2 GB | High | B57 |
Q6_K | 6 | 54.9 GB | High | B58 |
Q8_0Best for your GPU | 8 | 71.7 GB | Very High | B58 |
F16 | 16 | 137.4 GB | Maximum | F0 |
Get started
Copy-paste commands to run DeepSeek LLM 67B on your machine.
Run
docker run --rm -it ghcr.io/ggerganov/llama.cpp:full \
--hf-repo "deepseek-ai/deepseek-llm-67b-chat" \
--hf-file "deepseek-llm-67b-chat-Q4_K_M.gguf" \
-c 4096 -ngl 99Upgrade options