Pixtral 12B needs ~11.9 GB VRAM. Intel Arc B580 12GB has 12.0 GB. With Q4_K_M quantization, expect ~32 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 with offload
Decode
32.1 tok/s
TTFT
6023 ms
Safe context
17K
Memory
11.9 GB / 12.0 GB
The raw memory story may look fine, but the software ecosystem is still a constraint here.
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.
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 | A | Tight fit | 32.1 tok/s | 3285 ms | 17K |
| Coding | A | Runs with offload | 32.1 tok/s | 6023 ms | 17K |
| Agentic Coding | B | Very compromised (needs ~1.2 GB host RAM) | 17.2 tok/s | 16389 ms | 17K |
| Reasoning | A | Runs with offload | 32.1 tok/s | 7118 ms | 17K |
| RAG | B | Very compromised (needs ~1.2 GB host RAM) | 17.2 tok/s | 20487 ms | 17K |
Inference speed
Estimated decode speed (tokens/sec) for Pixtral 12B at Q4_K_M across popular GPUs and Apple Silicon, using the fastest local runtime per device. Fastest is RTX 5090 32GB at ~168 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? |
|---|---|---|---|---|
 RTX 5090 32GB | 32 GB | Q4_K_M | 168.0 | Fits |
| RTX 4090 24GB | 24 GB | Q4_K_M | 112.5 | Fits |
RX 7900 XTX 24GB | 24 GB | Q4_K_M | 101.5 | Fits |
| RTX 3090 24GB | 24 GB | Q4_K_M | 96.2 | Fits |
| RTX 4080 Super 16GB | 16 GB | Q4_K_M | 94.2 | Fits |
Mac Studio M3 Ultra 256GB | 256 GB | Q4_K_M | 81.8 | Fits |
Mac Studio M2 Ultra 128GB | 128 GB | Q4_K_M | 68.1 | Fits |
Mac Studio M1 Ultra 128GB | 128 GB | Q4_K_M | 64.6 | Fits |
| RTX 4070 12GB | 12 GB | Q4_K_M | 58.3 | Offloads |
MacBook Pro M4 Max 128GB | 128 GB | Q4_K_M | 44.5 | Fits |
MacBook Pro M4 Max 64GB | 64 GB | Q4_K_M | 44.5 | Fits |
MacBook Pro M3 Max 64GB | 64 GB | Q4_K_M | 35.2 | Fits |
| RTX 3060 12GB | 12 GB | Q4_K_M | 34.2 | Offloads |
MacBook Pro M1 Max 64GB | 64 GB | Q4_K_M | 32.3 | Fits |
MacBook Pro M4 Pro 48GB | 48 GB | Q4_K_M | 27.2 | Fits |
| RTX 4060 8GB | 8 GB | Q4_K_M | 9.7 | 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 Pixtral 12B (12B params) fits at each quantization level on Intel Arc B580 12GB (12.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 4.7 GB | Low | A75 |
Q3_K_S | 3 | 5.9 GB | Low | A76 |
NVFP4 | 4 | 6.7 GB | Medium | A76 |
Q4_K_M | 4 | 7.3 GB | Medium | A75 |
Q5_K_MBest for your GPU | 5 | 8.6 GB | High | A75 |
Q6_K | 6 | 9.8 GB | High | F0 |
Q8_0 | 8 | 12.8 GB | Very High | F0 |
F16 | 16 | 24.6 GB | Maximum | F0 |
Copy-paste commands to run Pixtral 12B on your machine.
Run
ollama run pixtralYour hardware
| Model | Params | Grade | Decode | Capabilities |
|---|---|---|---|---|
 Qwen 3 14B | 14B | A | 17.8 tok/s | |
 Phi-4-reasoning-plus 14B | 14.7B | A | 14.4 tok/s | |
 Ministral 3 14B | 14B | A | 17.7 tok/s | |
| Phi-4 14B | 14B | B | 16.1 tok/s | |
| Qwen 2.5 14B | 14B | B | 16.5 tok/s |
Yes, Intel Arc B580 12GB can run Pixtral 12B with a A grade (Runs with offload). Expected decode speed: 32.1 tok/s.
Pixtral 12B (12B parameters) requires approximately 11.9 GB of memory with Q4_K_M quantization.
The recommended quantization for Pixtral 12B is Q4_K_M, which balances quality and memory efficiency.
On Intel Arc B580 12GB, Pixtral 12B achieves approximately 32.1 tokens per second decode speed with a time-to-first-token of 6023ms using Q4_K_M quantization.
For coding workloads, Pixtral 12B on Intel Arc B580 12GB receives a A grade with 32.1 tok/s and 17K context.
On Intel Arc B580 12GB, Pixtral 12B can safely use up to 17K tokens of context. The model's official context limit is 131K, but available memory constrains the safe maximum.
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.
Paste this snippet into any page to show a live fit card.
<iframe src="https://willitrunai.com/embed/pixtral-12b-on-arc-b580-12gb" width="400" height="180" frameborder="0" style="border:none;border-radius:12px;overflow:hidden;" title="Will It Run AI — fit result"></iframe>
Preview: