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

RX 9060 XT 8 GB for local AI

Which image and video models run on the RX 9060 XT 8 GB, which file to download for each, and how much VRAM they need.

AMD8 GB GDDR6 · 128-bit · 320 GB/s · RDNA 4Data 2026-09-25
VRAM8 GB
MemoryGDDR6
Bus128-bit
Bandwidth320 GB/s
ArchitectureRDNA 4
Launched2025-06
Launch price$299
Runs well8 of 49

Specs: www.amd.com · launch: www.amd.com

01What runs on it

ModelVerdictBest fileSizeNeeded
image models
FLUX.1 [dev]TightQ3_K_S5.2 GB7.5 GB
FLUX.1 [schnell]TightQ3_K_S5.2 GB7.5 GB
FLUX.1 Kontext [dev]TightQ3_K_M5.4 GB8.0 GB
FLUX.1 Krea [dev]TightQ3_K_M5.4 GB7.7 GB
FLUX.1 Fill [dev]TightQ3_K_S5.2 GB7.8 GB
FLUX.2 [dev]Not practicalQ2_K12.9 GB16.2 GB
FLUX.2 [klein] 9BTightQ3_K_M4.8 GB7.1 GB
FLUX.2 [klein] 4BRuns wellFP84.1 GB5.9 GB
Krea 2 (Turbo)TightQ2_K4.9 GB7.5 GB
Qwen-ImageOffload onlyQ2_K7.1 GB9.9 GB
Qwen-Image-Edit (2511)Offload onlyQ4_K_M13.2 GB16.2 GB
Qwen-Image 2.1RunsQ5_K_M5.0 GB7.3 GB
Z-Image TurboRunsQ6_K5.9 GB7.9 GB
Z-Image (base)RunsQ5_K_M5.6 GB7.6 GB
Ideogram 4Offload onlyQ4_16.2 GB14.7 GB
Boogu-Image (Turbo)Offload onlyQ4_17.4 GB10.0 GB
ERNIE-Image (Turbo)RunsQ4_K_M5.0 GB7.3 GB
HiDream-O1-ImageOffload onlyFP88.1 GB10.9 GB
Mage-Flow (Microsoft)Runs wellINT84.2 GB6.0 GB
Lumina Image 2.0Runs well16-bit5.2 GB7.0 GB
HiDream-I1 (Full)Offload onlyQ2_K6.6 GB9.4 GB
HiDream-I1 (Dev)Offload onlyQ2_K6.6 GB9.4 GB
Stable Diffusion 3.5 LargeRunsQ4_15.3 GB7.6 GB
Stable Diffusion 3.5 MediumRuns well16-bit5.1 GB6.9 GB
Chroma1-HDRunsQ4_K_M5.6 GB7.9 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 2.1Offload onlyQ4_K_M11.3 GB14.6 GB
video models
Wan 2.1 T2V 14BOffload onlyQ4_K_M10.1 GB14.4 GB
Wan 2.1 T2V 1.3BRuns well16-bit2.8 GB5.6 GB
Wan 2.1 I2V 14B 480POffload onlyQ4_K_M11.3 GB15.6 GB
Wan 2.1 I2V 14B 720POffload onlyQ4_K_M11.3 GB18.1 GB
Wan 2.1 VACE 14BOffload onlyQ4_K_M11.6 GB16.4 GB
Wan 2.2 T2V A14BOffload onlyQ2_K5.3 GB9.6 GB
Wan 2.2 I2V A14BOffload onlyQ2_K5.3 GB9.6 GB
Wan 2.2 TI2V 5BRunsQ5_K_M3.8 GB7.6 GB
Wan 2.2 Animate 14BOffload onlyQ4_K_M11.5 GB16.8 GB
Wan Animate 2 (14B)Offload onlyQ4_K_M11.3 GB16.6 GB
Wan 2.2 S2V 14BOffload onlyQ4_K_M13.9 GB18.7 GB
SCAIL-2 (character animation)Offload onlyQ4_K_M11.5 GB16.8 GB
HunyuanVideo (13B, original)Offload onlyQ4_K_M7.9 GB12.2 GB
LTX-Video 13B (0.9.8)Offload onlyQ2_K4.7 GB9.0 GB
HunyuanVideo 1.5Offload onlyQ4_K_M5.1 GB9.4 GB
LTX-2 (19B)Offload onlyQ4_K_M12.8 GB17.6 GB
LTX-2.3 (22B)Offload onlyQ4_K_M14.3 GB19.1 GB
LTX-2.5 (22B)Offload onlyQ4_K_M15.1 GB19.9 GB
MiniMax H3 (33B)Not practicalQ3_K_M15.6 GB21.4 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

ComfyUI supports AMD GPUs on Windows officially since v0.7.0 (January 2026), through ROCm. Memory works the same as on NVIDIA, so the fit verdicts apply. Some custom nodes are NVIDIA-only. RX 9000 (RDNA 4) GPUs have FP8 hardware, and ComfyUI computes FP8 natively on them with PyTorch 2.7+ and ROCm 6.4+.

03Measured and reported results

No measured results yet. Send yours.