Qwen 2.5 Math 72B needs ~62.5 GB VRAM. Gaudi 3 128GB has 128.0 GB. With Q4_K_M quantization, expect ~64 tok/s.
Operating mode
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
Fit status
Runs well
Decode
64.1 tok/s
TTFT
3019 ms
Safe context
4K
Memory
62.5 GB / 128.0 GB
The raw memory story may look fine, but the software ecosystem is still a constraint here.
Runtime ecosystem is narrower than CUDA
Intel GPUs can look attractive on memory per dollar, but local AI tooling, kernels, and model coverage are still broader and easier on CUDA today.
Prefer CUDA if you want the path of least resistance
If your goal is maximum runtime coverage, easier troubleshooting, and better support for new local AI releases, CUDA is usually still the safer upgrade path.
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Chat | B | Runs well | 64.1 tok/s | 1647 ms | 4K |
| Coding | B | Runs well | 64.1 tok/s | 3019 ms | 4K |
| Agentic Coding | B | Runs well | 64.1 tok/s | 4391 ms | 4K |
| Reasoning | B | Runs well | 64.1 tok/s | 3568 ms | 4K |
| RAG | B | Runs well | 64.1 tok/s | 5489 ms | 4K |
How Qwen 2.5 Math 72B (72B params) fits at each quantization level on Gaudi 3 128GB (128.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 28.1 GB | Low | C54 |
Q3_K_S | 3 | 35.3 GB | Low | B55 |
NVFP4 | 4 |
Copy-paste commands to run Qwen 2.5 Math 72B on your machine.
Run
docker run --rm -it ghcr.io/ggerganov/llama.cpp:full \
--hf-repo "Qwen/Qwen2.5-Math-72B-Instruct" \
--hf-file "Qwen2.5-Math-72B-Instruct-Q4_K_M.gguf" \
-c 4096 -ngl 99Upgrade options
NVIDIA H200 141GBBudget pickRaises estimated decode speed by about 56%.
Moves you onto CUDA, which still has the broadest local-AI runtime coverage.
This is not only a hardware jump. It also gives you a cleaner runtime ecosystem for local LLM tooling.
~$30,000 MSRP
NVIDIA H200 PCIe 141GBBest valueRaises estimated decode speed by about 56%.
Moves you onto CUDA, which still has the broadest local-AI runtime coverage.
This is not only a hardware jump. It also gives you a cleaner runtime ecosystem for local LLM tooling.
~$30,000 MSRP
Yes, Gaudi 3 128GB can run Qwen 2.5 Math 72B with a B grade (Runs well). Expected decode speed: 64.1 tok/s.
Qwen 2.5 Math 72B (72B parameters) requires approximately 62.5 GB of memory with Q4_K_M quantization.
The recommended quantization for Qwen 2.5 Math 72B is Q4_K_M, which balances quality and memory efficiency.
On Gaudi 3 128GB, Qwen 2.5 Math 72B achieves approximately 64.1 tokens per second decode speed with a time-to-first-token of 3019ms using Q4_K_M quantization.
For coding workloads, Qwen 2.5 Math 72B on Gaudi 3 128GB receives a B grade with 64.1 tok/s and 4K context.
On Gaudi 3 128GB, Qwen 2.5 Math 72B can safely use up to 4K tokens of context. The model's official context limit is 4K, but available memory constrains the safe maximum.
Paste this snippet into any page to show a live fit card.
<iframe src="https://willitrunai.com/embed/qwen-2.5-math-72b-on-gaudi-3-128gb" width="400" height="180" frameborder="0" style="border:none;border-radius:12px;overflow:hidden;" title="Will It Run AI — fit result"></iframe>
Preview:
40.3 GB |
| Medium |
| B56 |
Q4_K_M | 4 | 43.9 GB | Medium | B57 |
Q5_K_M | 5 | 51.8 GB | High | B58 |
Q6_K | 6 | 59.0 GB | High | B59 |
Q8_0Best for your GPU | 8 | 77.0 GB | Very High | B61 |
F16 | 16 | 147.6 GB | Maximum | F0 |
Prefer CUDA if you want the path of least resistance. If your goal is maximum runtime coverage, easier troubleshooting, and better support for new local AI releases, CUDA is usually still the safer upgrade path.
Often yes, if your goal is the easiest setup and the widest runtime support. Intel can offer attractive memory capacity, but CUDA still tends to win on tooling maturity, guides, kernels, and model coverage for local AI.