What is the largest LLM the Intel Arc B580 can run?

Qwen3 14B, at Q4_K_M using about 10.9 GB of the 12 GB available, at roughly 30 tokens per second (8K context).

How many local AI models run well on the Intel Arc B580?

27 of the 69 models we track run fully in its 12 GB of VRAM at Q4_K_M or better.

Generation speed follows memory bandwidth. With 456 GB/s, expect about 32 tokens per second on an 8B model.

Every model we track, checked against the Intel Arc B580 at 8K context. The largest that runs well is Qwen3 14B.

VRAM12 GBIntel

Bandwidth456GB/s

Runs well27/69models at Q4_K_M or better

Fastest model~192tok/s · Gemma 4 26B-A4B

Bars show memory used out of 12 GB. Pick a row for the full breakdown.

NameMemory usedBest quantSpeed

NameMemory usedBest quantSpeed

NameMemory usedBest quantSpeed

Run it in the cloud

Rent a cloud GPU by the hour. No upfront cost, stop any time.