Qwen2.5 VL 32B Instruct on M3 Ultra 96 GB: VRAM, speed and fit
Qwen2.5 VL 32B Instruct at 8-bit runs on the M3 Ultra 96 GB with 125K of context at about 14 tokens a second.
Text generation · Chat · Ollama / LM Studio
Runs comfortably · on M3 Ultra 96 GB · 8-bit (Q8_0) · Ollama / LM Studio · Chat
- GPU memory
- 89.0 GB
- of 96.0 GB
- Context
- 125K
- auto
- Speed
- 14 tok/s
- 7.5 when full
- First reply
- 6.3 s
- 2K prompt
Where the memory goes
- Model
- 33.1 GB
- Context
- 31.3 GB
- Overhead
- 713 MB
- OS
- 24.0 GB
- Free
- 7.0 GB
- Memory
- 89.0 GB of 96.0 GB
- Context
- 125K tokens · auto
- Speed
- 14 tok/s · 7.5 when full
- First reply
- 6.3 s for a 2K prompt
- Running cost
- $3.55/mo at home · $122/mo on a cloud 4× RTX A4500
Validation: Not validated
| Check | Target | Result | Status |
|---|---|---|---|
| Fits in GPU memory | 72.0 GB usable | 65.0 GB needed | Pass |
| Headroom | 5% free | 7.0 GB free (10%) | Pass |
| Context for chat | 16K or more | 125K | Pass |
| Writing speed | 15 tok/s or more | 14 tok/s | Check |
| First reply | under 3 s, 2K prompt | 6.3 s | Fails |
Targets: a reply you can read along with, a first reply the use case tolerates, and 5% of memory left free. Calculated from published specs, not benchmarked.
The setup
- Model
- Qwen2.5 VL 32B Instruct
- Parameters
- 33.5B
- Precision
- 8-bit (Q8_0)
- Context
- 125K, auto
- Chats at once
- 1
- Use
- Chat
- Runs in
- Ollama / LM Studio
- Hardware
- M3 Ultra 96 GB
- GPU can use
- 72.0 GB of 96.0 GB
Running cost
- At home
- $3.55/mo
- 162 W at the wall
- Renting instead
- $122/mo
- 4× RTX A4500 at $1.00/h
4 h a day, power at $0.18 per kWh, cloud at RunPod on-demand prices as of Nov 21, 2026.
How to run it
- llama.cpp
llama-server -hf bartowski/Qwen_Qwen2.5-VL-32B-Instruct-GGUF- vLLM
vllm serve Qwen/Qwen2.5-VL-32B-Instruct
Qwen2.5 VL 32B Instruct on similar machines
- Mac mini M4 Pro 48 GBRuns
- Mac mini M4 Pro 64 GBRuns
- M4 Max 128 GBRuns
- M5 Max 128 GBRuns
- M3 Ultra 256 GBRuns
- M3 Ultra 512 GBRuns
- M5 Ultra 512 GBRuns
- Ryzen AI Max+ 395 64 GBRuns
All machines: Qwen2.5 VL 32B Instruct VRAM requirements. Everything that runs here: what can the M3 Ultra 96 GB run?
Other models that run on M3 Ultra 96 GB
Questions
Can Qwen2.5 VL 32B Instruct run on M3 Ultra 96 GB?
Qwen2.5 VL 32B Instruct at 8-bit runs on the M3 Ultra 96 GB with 125K of context at about 14 tokens a second.
Report VR-0XWRCAQ, calculated Nov 22, 2026 from published specs, not benchmarked.