GPUs & Macs › Intel Arc Pro B60
Intel Arc Pro B60: local LLM speed
Llama 2 7B (Q4_0) generates 70.5 tokens/s on the Intel Arc Pro B60 (Vulkan), and processes prompts at 275 tokens/s.
Measured results
| Model | Generate tok/s | Prompt tok/s | Quant | Backend | Test | Source |
|---|---|---|---|---|---|---|
| Llama 2 7B | 70.5 | 275 | Q4_0 | Vulkan FA on | tg128 | llama.cpp #10879 details |
| Llama 2 7B | 68.5 | 522 | Q4_0 | Vulkan FA off | tg128 | llama.cpp #10879 details |
What fits on the Intel Arc Pro B60
Estimates for this chip are scaled from a measured run: Llama 2 7B Q4_0 at 70.5 tok/s (Vulkan, source). Fit assumes the whole model in GPU memory with ~8K context.
| Model | Best quant that fits | File | Expected tok/s |
|---|---|---|---|
| Qwen3.8-27B | UD-Q6_K | 22.0 GB | 12.3 estimate |
| gemma-4-12B-it | Q8_0 | 12.7 GB | 21.3 estimate |
| Qwen3.5-9B | Q8_0 | 9.5 GB | 28.3 estimate |
| Qwen3.5-4B | Q8_0 | 4.5 GB | 60.1 estimate |
| Qwen3.6-35B-A3B | UD-Q4_K_S | 20.9 GB | 75.3 estimate |
| gemma-4-26B-A4B-it | UD-Q5_K_M | 21.2 GB | 41.4 estimate |
| MiMo-V2.6-Distill-Qwen-9B | Q8_0 | 9.5 GB | 28.2 estimate |
| gemma-4-E4B-it | Q8_0 | 8.2 GB | 32.9 estimate |
| Muse-Glimmer-30B | UD-Q5_K_M | 19.2 GB | 14.0 estimate |
| Qwen3-4B | Q8_0 | 4.3 GB | 63.0 estimate |
| gemma-4-31B-it | Q5_K_M | 21.7 GB | 12.4 estimate |
| gemma-4-26B-A4B-it-qat-q4_0-unquantized | UD-Q4_K_XL | 14.2 GB | 61.5 estimate |
| gemma-4-12B-it-qat-q4_0-unquantized | UD-Q4_K_XL | 6.7 GB | 40.1 estimate |
| gemma-3-1b-it | Q8_0 | 1.1 GB | 225 estimate |