Datasheet for local AI49 models98 GPUsData read 2026-09-25
No adsNo tracking
Check my GPU →

Arc B580 12 GB for local AI

Which image and video models run on the Arc B580 12 GB, which file to download for each, and how much VRAM they need.

Intel12 GB GDDR6 · 192-bit · 456 GB/s · BattlemageData 2026-09-25
VRAM12 GB
MemoryGDDR6
Bus192-bit
Bandwidth456 GB/s
ArchitectureBattlemage
Launched2024-12
Launch price$249
Runs well17 of 49

Specs: www.intel.com · launch: www.intel.com

01What runs on it

ModelVerdictBest fileSizeNeeded
image models
FLUX.1 [dev]RunsQ5_K_S8.3 GB10.6 GB
FLUX.1 [schnell]RunsQ5_K_S8.3 GB10.6 GB
FLUX.1 Kontext [dev]RunsQ5_K_M8.4 GB11.0 GB
FLUX.1 Krea [dev]RunsQ5_K_M8.4 GB10.7 GB
FLUX.1 Fill [dev]RunsQ5_K_S8.3 GB10.9 GB
FLUX.2 [dev]Offload onlyQ4_K_M20.1 GB23.4 GB
FLUX.2 [klein] 9BRuns wellFP89.4 GB11.7 GB
FLUX.2 [klein] 4BRuns well16-bit7.8 GB9.6 GB
Krea 2 (Turbo)RunsQ5_K_M8.9 GB11.5 GB
Qwen-ImageTightQ2_K7.1 GB9.9 GB
Qwen-Image-Edit (2511)TightQ2_K7.5 GB10.5 GB
Qwen-Image 2.1Runs wellQ8_07.6 GB9.9 GB
Z-Image TurboRuns wellQ8_07.2 GB9.2 GB
Z-Image (base)Runs wellQ8_07.2 GB9.2 GB
Ideogram 4Offload onlyQ4_16.2 GB14.7 GB
Boogu-Image (Turbo)RunsQ5_18.6 GB11.2 GB
ERNIE-Image (Turbo)Runs wellQ8_08.7 GB11.0 GB
HiDream-O1-ImageRuns wellFP88.1 GB10.9 GB
Mage-Flow (Microsoft)Runs well16-bit8.2 GB10.0 GB
Lumina Image 2.0Runs well16-bit5.2 GB7.0 GB
HiDream-I1 (Full)TightQ3_K_M8.8 GB11.6 GB
HiDream-I1 (Dev)TightQ3_K_M8.8 GB11.6 GB
Stable Diffusion 3.5 LargeRuns wellQ8_08.8 GB11.1 GB
Stable Diffusion 3.5 MediumRuns well16-bit5.1 GB6.9 GB
Chroma1-HDRuns wellFP89.2 GB11.5 GB
SDXL 1.0Runs well16-bit6.9 GB7.1 GB
Illustrious XL / Pony (SDXL anime)Runs well16-bit6.9 GB7.1 GB
Stable Diffusion 1.5Runs well16-bit2.1 GB3.3 GB
HunyuanImage 2.1TightQ2_K7.3 GB10.6 GB
video models
Wan 2.1 T2V 14BTightQ3_K_M7.6 GB11.9 GB
Wan 2.1 T2V 1.3BRuns well16-bit2.8 GB5.6 GB
Wan 2.1 I2V 14B 480POffload onlyQ3_K_M8.6 GB12.9 GB
Wan 2.1 I2V 14B 720POffload onlyQ4_K_M11.3 GB18.1 GB
Wan 2.1 VACE 14BOffload onlyQ3_K_S7.8 GB12.6 GB
Wan 2.2 T2V A14BTightQ3_K_M7.2 GB11.5 GB
Wan 2.2 I2V A14BTightQ3_K_M7.2 GB11.5 GB
Wan 2.2 TI2V 5BRuns wellQ8_05.4 GB9.2 GB
Wan 2.2 Animate 14BTightQ2_K6.5 GB11.8 GB
Wan Animate 2 (14B)TightQ2_K6.5 GB11.8 GB
Wan 2.2 S2V 14BOffload onlyQ4_K_M13.9 GB18.7 GB
SCAIL-2 (character animation)Offload onlyQ2_K7.3 GB12.6 GB
HunyuanVideo (13B, original)TightQ3_K_M6.2 GB10.5 GB
LTX-Video 13B (0.9.8)TightQ3_K_M6.5 GB10.8 GB
HunyuanVideo 1.5RunsQ6_K7.0 GB11.3 GB
LTX-2 (19B)Offload onlyQ2_K8.1 GB12.9 GB
LTX-2.3 (22B)Offload onlyQ2_K8.3 GB13.1 GB
LTX-2.5 (22B)Offload onlyQ2_K8.8 GB13.6 GB
MiniMax H3 (33B)Offload onlyQ4_K_M19.9 GB25.7 GB
MiniMax H3 PrunedOffload onlyQ4_K_M11.6 GB17.4 GB

Calculated from real file sizes plus working memory. How the numbers work.

02Good to know

Intel Arc GPUs run ComfyUI through PyTorch's native XPU support on Windows 11 and Linux. Memory works the same as on NVIDIA, so the fit verdicts apply. FP8 files save memory but give no speed-up, and some custom nodes are NVIDIA-only.

03Measured and reported results

LabelModelSetupResultPeak VRAMDateSource
reportedFLUX.1 devfp8 · 1024x1024 · 20 steps
“Flux1 dev fp8 (20step): GOOD (35s)”
Intel Arc B580 12GB, PyTorch 2.8 XPU in Docker (YanWenKun); single 1024x1024 image, pre-warmed inference only (model load excluded)
35 s / image—2025-08-03github.com →
reportedFLUX.1 KreaFlux1 Krea dev (precision not stated) · 1024x1024 · 20 steps
“Flux1 Krea dev (20step): OK (46s)”
Intel Arc B580 12GB, PyTorch 2.8 XPU in Docker (YanWenKun); single 1024x1024 image, pre-warmed inference only (model load excluded)
46 s / image—2025-08-03github.com →
reportedFLUX.1 schnellFlux1 schnell (template) · 1024x1024 · 4 steps
“Flux1 schnell (4step): GOOD (8s)”
Intel Arc B580 12GB, PyTorch 2.8 XPU in Docker (YanWenKun); single 1024x1024 image, pre-warmed inference only (model load excluded)
8 s / image—2025-08-03github.com →
reportedSD 3.5 Largefp8 · 1024x1024 · 20 steps
“SD 3.5 large fp8 (20step): GOOD (26s)”
Intel Arc B580 12GB, PyTorch 2.8 XPU in Docker (YanWenKun); single 1024x1024 image, pre-warmed inference only (model load excluded)
26 s / image—2025-08-03github.com →
reportedSDXLsd_xl_base_1.0 · 1024x1024 · 20 steps
“100%|██| 20/20 [00:05<00:00, 3.96it/s]”
ComfyUI 'GPU Benchmark' thread: default workflow, SDXL 1.0 base, 1024x1024, seed 1, second run; Intel B580 Steel Legend OC 12 GB
3.96 it/s—2025-05-12github.com →
reportedWan 2.2 5BWan 2.2 5B T2V (template) · 640x352
“Wan 2.2 5B Text to Video (640x352, 121 frames): FAST (60s, very poor quality)”
Arc B580 12GB, PyTorch 2.8 XPU; same post: 960x544/121f = 181s, 1280x704 OOM; steps not stated
60 s / clip (121 frames)—2025-08-03github.com →

Reported results are other people's numbers, copied as published, with a link. Settings, drivers and ComfyUI versions differ, so compare them with care. Send yours.