Intel GPU

Intel Arc B580

The Intel Arc B580 has 12GB of VRAM and 456 GB/s of memory bandwidth. Its sweet spot tops out around Mistral-Nemo-Instruct-2407 — the largest model that loads cleanly at a quality quant with llama.cpp at 8K context.

Intel verdicts are beta: fit is physics and holds, but ROCm/Vulkan speed tiers aren't calibrated yet — treat tiers as provisional until real measurements land.

VRAM: 12 GB
Memory bandwidth: 456 GB/s
TDP: 190 W
MSRP: $249
Backends: Vulkan

Also good for gaming — 1080p-class (estimated from FP32 compute, not a game benchmark)

Specs as of 2026-07-08 · source

Check price on Amazon

Intel Arc B580 for local AI — common questions

How much VRAM does the Intel Arc B580 have?: The Intel Arc B580 has 12GB of VRAM with 456 GB/s of memory bandwidth. VRAM sets which models fit; bandwidth sets how fast they feel once they do.
Is the Intel Arc B580 good for running AI models locally?: Yes, within limits. 12GB comfortably runs small and mid-size models at a quality quant, but the largest open models will not fit. The largest model it loads cleanly is Mistral-Nemo-Instruct-2407, at a quality quant with llama.cpp at 8K context.
What runtimes work with the Intel Arc B580?: It runs on Vulkan, which covers Ollama, llama.cpp and LM Studio. Runtime choice affects speed and features, not whether a model fits.

Affiliate link — we may earn a commission. Verdicts are computed before any link is attached.

What an Intel Arc B580 can run

Every model we track, graded at its best clean-fit quant. Click any verdict for the full breakdown.

Model	Best-quant verdict · llama.cpp · 8K context
Qwen3-8B	Fits · B · Q6_Kinteractivebeta
gemma-4-26B-A4B-it	Tight · C · MXFP4_MOEinteractive · offloadbeta
Qwen2.5-7B-Instruct	Fits · B · Q8_0interactivebeta
Qwen2.5-1.5B-Instruct	Fits · A · Q8_0interactivebeta
Llama-3.2-1B-Instruct	Fits · A · Q8_0interactivebeta
Llama-3.1-8B-Instruct	Fits · B · Q8_0interactivebeta
DeepSeek-R1	Won't fitbeta
Qwen3.5-9B	Fits · B · Q6_Kinteractivebeta
gpt-oss-20b	Tight · C · MXFP4interactive · offloadbeta
Qwen3.6-35B-A3B	Tight · C · MXFP4_MOEinteractive · offloadbeta
Qwen2.5-3B-Instruct	Fits · A · Q8_0interactivebeta
Qwen3-32B	Tight · C · Q4_K_Mpainful · offloadbeta
Qwen3.6-27B	Tight · C · Q4_K_Mpainful · offloadbeta
Qwen2.5-0.5B-Instruct	Fits · A · Q8_0interactivebeta
Mistral-7B-Instruct-v0.3	Fits · B · Q8_0interactivebeta
gpt-oss-120b	Tight · C · MXFP4interactive · offloadbeta
Qwen3-14B	Tight · C · Q4_K_Minteractivebeta
Qwen3-30B-A3B	Tight · C · Q4_K_Minteractive · offloadbeta
Qwen2.5-32B-Instruct	Tight · C · Q4_K_Mpainful · offloadbeta
DeepSeek-V4-Flash	Won't fitbeta
Qwen2.5-14B-Instruct	Tight · C · Q4_K_Minteractivebeta
Llama-3.2-3B-Instruct	Fits · A · Q8_0interactivebeta
gemma-3-4b-it	Fits · A · Q8_0interactivebeta
gemma-3-12b-it	Tight · C · Q4_K_Minteractivebeta
Llama-3.1-70B-Instruct	Tight · C · Q4_K_Mpainful · offloadbeta
Phi-3.5-mini-instruct	Fits · A · Q8_0interactivebeta
Phi-4	Tight · C · Q4_K_Minteractivebeta
DeepSeek-R1-Distill-Qwen-32B	Tight · C · Q4_K_Mpainful · offloadbeta
gemma-3-27b-it	Tight · C · Q4_K_Mpainful · offloadbeta
Llama-3.3-70B-Instruct	Tight · C · Q4_K_Mpainful · offloadbeta
Qwen2.5-72B-Instruct	Tight · C · Q4_K_Mpainful · offloadbeta
gemma-2-2b-it	Fits · A · Q8_0interactivebeta
DeepSeek-R1-Distill-Qwen-14B	Tight · C · Q4_K_Minteractivebeta
DeepSeek-R1-Distill-Llama-8B	Fits · B · Q8_0interactivebeta
DeepSeek-R1-Distill-Qwen-7B	Fits · B · Q8_0interactivebeta
gemma-2-9b-it	Fits · B · Q6_Kinteractivebeta
Mistral-Nemo-Instruct-2407	Fits · B · Q4_K_Minteractivebeta
QwQ-32B	Tight · C · Q4_K_Mpainful · offloadbeta
Mistral-Small-24B-Instruct-2501	Tight · C · Q4_K_Musable · offloadbeta
gemma-2-27b-it	Tight · C · Q4_K_Mpainful · offloadbeta

Compare Intel Arc B580

Have a specific model in mind? Check it against the Intel Arc B580 at your own quant and context length.

Check with the Intel Arc B580 preselected →

Verdicts computed from published specs and measured GGUF sizes — how the math works.