Mixtral 8x7B needs ~28.3 GB VRAM. Intel Arc Pro B60 24GB has 24.0 GB. With Q3_K_S quantization, expect ~11 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
9.9 GB over capacity — needs offload or smaller quantization
Fit status
Too heavy
Decode
6.7 tok/s
TTFT
28955 ms
Safe context
4K
Memory
33.9 GB / 24.0 GB
Offload
30%
It fits through host-memory offload, and offload is the main reason performance drops.
CPU or host-memory offload is active
About 20% of the working set spills out of accelerator memory, which usually hurts latency and sustained decode throughput.
Very little memory headroom
You can run the model, but there is not much room left for longer context, bigger batches, extra apps, or future model updates.
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.
Remove offload with more accelerator memory
Prioritize a GPU or unified-memory tier that fits the whole model natively. Removing offload usually helps more than small compute gains.
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.
Buy headroom, not only minimum fit
A slightly larger memory tier gives you safer context growth and makes the recommendation more future-proof.
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Chat | F | Too heavy | 7.1 tok/s | 14880 ms | 4K |
| Coding | F | Too heavy | 6.7 tok/s | 28955 ms | 4K |
| Agentic Coding | F | Too heavy | 6.0 tok/s | 47211 ms | 4K |
| Reasoning | F | Too heavy | 6.7 tok/s | 34219 ms | 4K |
| RAG | F | Too heavy | 6.0 tok/s | 59014 ms | 4K |
Inference speed
Estimated decode speed (tokens/sec) for Mixtral 8x7B at Q4_K_M across popular GPUs and Apple Silicon, including multi-GPU rigs, using the fastest local runtime per device. Fastest is 2× RX 7900 XTX 24GB at ~85 tok/s. Speed is memory-bandwidth bound, so cards that fit the whole model in VRAM run far faster than ones that offload to system RAM.
| GPU / Mac | Memory | Quant | Speed (tok/s) | Fits? |
|---|---|---|---|---|
2× RX 7900 XTX 24GB | 48 GB | Q4_K_M | 84.5 | Fits |
 2× RTX 4090 24GB | 48 GB | Q4_K_M | 77.1 | Fits |
| 2× RTX 3090 24GB | 48 GB | Q4_K_M | 66.0 | Fits |
| 4× RTX 3060 12GB | 48 GB | Q4_K_M | 58.1 | Fits |
| RTX 5090 32GB | 32 GB | Q4_K_M | 54.9 | Heavy offload |
Mac Studio M3 Ultra 256GB | 256 GB | Q4_K_M | 40.1 | Fits |
Mac Studio M2 Ultra 128GB | 128 GB | Q4_K_M | 33.4 | Fits |
Mac Studio M1 Ultra 128GB | 128 GB | Q4_K_M | 31.6 | Fits |
MacBook Pro M4 Max 128GB | 128 GB | Q4_K_M | 24.7 | Fits |
MacBook Pro M4 Max 64GB | 64 GB | Q4_K_M | 24.7 | Tight |
| RTX 4090 24GB | 24 GB | Q4_K_M | 19.6 | Too big |
RX 7900 XTX 24GB | 24 GB | Q4_K_M | 18.0 | Too big |
MacBook Pro M3 Max 64GB | 64 GB | Q4_K_M | 17.3 | Tight |
| RTX 3090 24GB | 24 GB | Q4_K_M | 16.7 | Too big |
MacBook Pro M1 Max 64GB | 64 GB | Q4_K_M | 15.8 | Tight |
MacBook Pro M4 Pro 48GB | 48 GB | Q4_K_M | 13.6 | Offloads |
| RTX 4080 Super 16GB | 16 GB | Q4_K_M | 7.0 | Too big |
| RTX 4070 12GB | 12 GB | Q4_K_M | 4.1 | Too big |
| RTX 3060 12GB | 12 GB | Q4_K_M | 2.6 | Too big |
| RTX 4060 8GB | 8 GB | Q4_K_M | 2.1 | Too big |
Estimates for single-stream decoding at Q4_K_M; real tokens/sec varies with prompt length, context, batch size, and runtime build. Prompt processing (prefill) is faster than the decode figures shown here.
How Mixtral 8x7B (47B params) fits at each quantization level on Intel Arc Pro B60 24GB (24.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 18.3 GB | Low | F0 |
Q3_K_S | 3 | 23.0 GB | Low | F0 |
NVFP4 | 4 | 26.3 GB | Medium | F0 |
Q4_K_M | 4 | 28.7 GB | Medium | F0 |
Q5_K_M | 5 | 33.8 GB | High | F0 |
Q6_K | 6 | 38.5 GB | High | F0 |
Q8_0 | 8 | 50.3 GB | Very High | F0 |
F16 | 16 | 96.4 GB | Maximum | F0 |
Copy-paste commands to run Mixtral 8x7B on your machine.
Run
ollama run mixtral升级选项
RTX PRO 5000 Blackwell 48GB最大飞跃Makes the model fit on the accelerator instead of staying completely out of reach.
Removes host-memory offload, which is usually the single biggest latency and throughput win.
~$4,999 MSRP
Intel Data Center GPU Max 1550 128GB经济之选Makes the model fit on the accelerator instead of staying completely out of reach.
Removes host-memory offload, which is usually the single biggest latency and throughput win.
~$15,000 MSRP
Gaudi 3 128GB最佳性价比Makes the model fit on the accelerator instead of staying completely out of reach.
Removes host-memory offload, which is usually the single biggest latency and throughput win.
~$15,000 MSRP
Yes, Intel Arc Pro B60 24GB can run Mixtral 8x7B at Q3_K_S quantization (Very compromised (needs ~3.5 GB host RAM)). The recommended Q4_K_M requires 33.9 GB which exceeds available memory, but at Q3_K_S it needs only 28.3 GB. Expected decode speed: 11.2 tok/s.
Mixtral 8x7B (47B parameters) requires approximately 33.9 GB at Q4_K_M quantization. On Intel Arc Pro B60 24GB, it fits at Q3_K_S using 28.3 GB.
The recommended quantization is Q4_K_M, but on Intel Arc Pro B60 24GB the best fitting quantization is Q3_K_S, which uses 28.3 GB.
On Intel Arc Pro B60 24GB, Mixtral 8x7B achieves approximately 11.2 tokens per second decode speed with a time-to-first-token of 17259ms using Q3_K_S quantization.
For coding workloads, Mixtral 8x7B on Intel Arc Pro B60 24GB receives a F grade with 6.7 tok/s and 4K context.
On Intel Arc Pro B60 24GB, Mixtral 8x7B can safely use up to 4K tokens of context at Q3_K_S quantization. The model's official context limit is 33K, but available memory constrains the safe maximum.
Remove offload with more accelerator memory. Prioritize a GPU or unified-memory tier that fits the whole model natively. Removing offload usually helps more than small compute gains.
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.
Paste this snippet into any page to show a live fit card.
<iframe src="https://willitrunai.com/embed/mixtral-8x7b-on-arc-pro-b60-24gb" width="400" height="180" frameborder="0" style="border:none;border-radius:12px;overflow:hidden;" title="Will It Run AI — fit result"></iframe>
Preview: