Datasheet No. 01 · Local AI on your own GPUCalculated from 851 real model files
Will it run
on your GPU?
FLUX, Qwen-Image, Krea 2, Wan, LTX, MiniMax H3 and 44 more. Which file to download, how much VRAM it really needs, and what your card can handle. Every number says where it came from.
I have and I want to run .
…
Loading the answer.
Memory map · RTX 5060 Ti 16 GB · FLUX.1 dev FP8 · Sampling14.2 GB / 16 GB
08 GB16 GB
Weights FP8 · 11.9 GBWorking memory · 1.5 GBReserve · 0.8 GBFree · 1.8 GB
50image & video models
99desktop & laptop GPUs
851model files, sized to the byte
2501answer pages
01The memory ruler
Every model on one scale. Drag the red line to your VRAM.
Each dot is one file of a model, placed at the VRAM it needs. Everything left of the line fits.
Your VRAM
16GB
0 of 50 models fit entirely.
Model
081624324048
Verdict · file
Image models
FLUX.1 dev12B · imageQ2_KQ3_K_SQ4_K_SQ5_K_SQ6_KFP8Q8_016-bitFLUX.1 schnell12B · imageQ2_KQ3_K_SQ4_K_SQ5_K_SQ6_KFP8Q8_016-bitFLUX.1 Kontext12B · imageQ2_KQ3_K_MQ4_K_MQ5_K_MQ6_KFP8Q8_016-bitFLUX.1 Krea12B · imageQ2_KQ3_K_MQ4_K_MQ5_K_MQ6_KFP8Q8_016-bitFLUX.1 Fill12B · imageQ3_K_SQ4_K_SQ5_K_SQ6_KQ8_016-bitFLUX.2 dev32B · imageQ2_KQ3_K_MQ4_K_MQ5_K_MQ6_KQ8_0FP8FLUX.2 klein 9B9B · imageQ2_KQ3_K_MQ4_K_MQ5_K_MQ6_KFP8Q8_016-bitFLUX.2 klein 4B4B · imageQ2_KQ3_K_MQ4_K_MQ5_K_MQ6_KFP8Q8_016-bitKrea 212.8B · imageQ2_KQ3_K_MQ4_K_MQ5_K_MQ6_KFP8INT8Q8_016-bitQwen-Image20B · imageQ2_KQ3_K_MQ4_K_MQ5_K_MQ6_KFP8Q8_016-bitQwen-Image-Edit20B · imageQ2_KQ3_K_MQ4_K_MQ5_K_MQ6_KFP8Q8_016-bitQwen-Image 2.17.1B · imageQ3_K_MQ4_K_MQ5_K_MQ6_KINT8Q8_016-bitZ-Image Turbo6B · imageQ2_KQ3_K_MQ4_K_MQ5_K_MQ6_KINT8Q8_016-bitZ-Image6.15B · imageQ2_KQ3_K_MQ4_K_MQ5_K_MQ6_KINT8Q8_016-bitIdeogram 49.3B · imageQ4_1Q5_1FP8INT8Q8_0Boogu-Image10.29B · imageQ4_1Q5_1FP8INT8Q8_016-bitERNIE-Image8.03B · imageQ2_KQ3_K_MQ4_K_MQ5_K_MQ6_KQ8_016-bitHiDream-O18.8B · imageFP816-bitMage-Flow4.12B · imageINT816-bitMing-Image6.15B · imageQ2_KQ3_K_MQ4_K_MQ5_K_MINT8Q6_KQ8_016-bitLumina 2.02.61B · imageQ2_KQ3_K_MQ4_K_MQ5_K_MQ6_KQ8_016-bitHiDream-I1 Full17B · imageQ2_KQ3_K_MQ4_K_MQ5_K_MQ6_KFP8Q8_016-bitHiDream-I117B · imageQ2_KQ3_K_MQ4_K_MQ5_K_MQ6_KFP8Q8_016-bitSD 3.5 Large8.1B · imageQ4_1Q5_1Q8_016-bitSD 3.5 Medium2.5B · imageQ3_K_MQ4_K_MQ5_K_MQ6_KQ8_016-bitChroma1-HD8.9B · imageQ2_KQ3_K_SQ4_K_MQ5_K_MQ6_KFP8Q8_016-bitSDXL3.5B · image16-bitIllustrious / Pony3.5B · imageQ2_KQ3_K_MQ4_K_MQ5_K_MQ6_KQ8_0FP816-bitSD 1.50.98B · image16-bitHunyuanImage 2.117B · imageQ2_KQ3_K_MQ4_K_MQ5_K_MQ6_KFP8Q8_016-bitVideo models
Wan 2.1 14B14B · videoQ3_K_MQ4_K_MQ5_K_MQ6_KFP8Q8_016-bitWan 2.1 1.3B1.3B · videoQ3_K_MQ4_K_MQ5_K_MQ6_KQ8_016-bitWan 2.1 I2V 480P16.4B · videoQ3_K_MQ4_K_MQ5_K_MQ6_KFP8Q8_016-bitWan 2.1 I2V 720P16.4B · videoQ3_K_MQ4_K_MQ5_K_MQ6_KFP8Q8_016-bitWan VACE 14B17.3B · videoQ3_K_SQ4_K_MQ5_K_MQ6_KQ8_016-bitWan 2.2 T2V14B · videoQ2_KQ3_K_MQ4_K_MQ5_K_MQ6_KFP8Q8_016-bitWan 2.2 I2V14B · videoQ2_KQ3_K_MQ4_K_MQ5_K_MQ6_KFP8Q8_016-bitWan 2.2 5B5B · videoQ2_KQ3_K_MQ4_K_MQ5_K_MQ6_KQ8_016-bitWan 2.2 Animate17.3B · videoQ2_KQ3_K_MQ4_K_MQ5_K_MQ6_KFP8INT8Q8_016-bitWan Animate 214B · videoQ2_KQ3_K_MQ4_K_MQ5_K_MQ6_KINT8Q8_016-bitWan 2.2 S2V16.3B · videoQ2_KQ3_K_MQ4_K_MQ5_K_MQ6_KFP8Q8_016-bitSCAIL-216B · videoQ2_KQ3_K_MQ4_K_MQ5_K_MQ6_KINT8FP8Q8_016-bitHunyuanVideo 13B13B · videoQ3_K_MQ4_K_MQ5_K_MQ6_KFP8Q8_016-bitLTX-Video 13B13B · videoQ2_KQ3_K_MQ4_K_MQ5_K_MQ6_KQ8_0FP816-bitHunyuanVideo 1.58.3B · videoQ4_K_MQ5_K_MQ6_KFP8Q8_016-bitLTX-219B · videoQ2_KQ3_K_MQ4_K_MQ5_K_MQ6_KQ8_016-bitLTX-2.322B · videoQ2_KQ3_K_MQ4_K_MQ5_K_MQ6_KQ8_016-bitLTX-2.522B · videoQ2_KQ3_K_MQ4_K_MQ5_K_MQ6_KQ8_016-bitMiniMax H333.1B · videoQ3_K_MQ4_K_MQ5_K_MQ6_KINT8Q8_0MiniMax H3 Prunedpruned 33B · videoQ3_K_MQ4_K_MQ5_K_MQ6_KFP8INT8Q8_016-bitA file of this model (VRAM it needs)Fits in your VRAMBest file that fitsYour VRAM
Q.01Can the RTX 4060 8 GB run FLUX.1 dev?Tight Q3_K_S · 7.5 GBQ.02Can the RTX 3060 12 GB run FLUX.1 dev?Runs Q5_K_S · 10.6 GBQ.03Can the RTX 5060 Ti 16 GB run Wan 2.2 I2V?Runs Q5_K_M · 15.1 GBQ.04Can the RTX 4090 24 GB run Qwen-Image?Runs well FP8 · 23.2 GBQ.05Can the RTX 3060 Ti 8 GB run Z-Image Turbo?Runs Q6_K · 7.9 GBQ.06Can the RTX 5090 32 GB run FLUX.2 dev?Runs Q6_K · 30.7 GBQ.07Can the RTX 2060 6 GB run SDXL?Offload only 16-bit · 7.1 GBQ.08Can the RTX 4070 Super 12 GB run Krea 2?Runs Q5_K_M · 11.5 GBQ.09Can the RTX 5090 32 GB run MiniMax H3 Pruned?Runs well FP8 · 26.8 GBQ.10Can the RTX 4060 Laptop 8 GB run Wan 2.2 5B?Runs Q5_K_M · 7.6 GBQ.11Can the RTX 5060 8 GB run Qwen-Image 2.1?Runs Q5_K_M · 7.3 GBQ.12Can the RTX 3060 Ti 8 GB run Illustrious / Pony?Runs well 16-bit · 7.1 GB
03Start from what you have
Browse by VRAM
The shaded part of each column is how many models fit at all.
04Tools and special cases
Plan the setting, the upgrade, the laptop
CalculatorBigger images, longer clips, LoRAs and ControlNetsSet resolution, frames and batch; see which file still fits →
UpgradeWhat would a new GPU change?Your card vs the one you want: what newly runs well →
TrainingLoRA training: how much VRAM49 figures from the trainers' own docs, with settings →
LaptopsLaptop GPUs have less VRAM than their names suggestThe RTX 4090 Laptop has 16 GB, not 24. 32 laptop GPUs →
FreeComfyUI workflows, tested on 16 GB15 ready workflows with file links and measured speed →
Compare44 GPU matchups, model by modelRTX 3090 vs 5060 Ti, 7900 XTX vs 4090… →
calculated
From real file sizes
Every model file is sized to the byte from Hugging Face. Add the model's working memory and a small system reserve, and you get the verdict of what fits entirely.
measured
On my own GPU
I make AI images in ComfyUI on an RTX 5060 Ti 16 GB. A script times each model and logs peak VRAM: 12 models timed so far. See the numbers or get the workflows I used.
reported
From other people
94 real-world results from GitHub, Hugging Face and first-hand blogs, copied as published with a link to each. See them all or send yours.
2026-09Qwen-Image 2.17.1B · image · fits from 6 GB2026-09Ming-Image 0.1 Design6.15B · image · fits from 8 GB2026-07Mage-Flow (Microsoft)4.12B · image · fits from 6 GB2026-07Wan Animate 2 (14B)14B · video · fits from 12 GB2026-07LTX-2.5 (22B)22B · video · fits from 14 GB2026-07MiniMax H3 (33B)33.1B · video · fits from 22 GB2026-07MiniMax H3 Prunedpruned 33B · video · fits from 15 GB2026-06Krea 2 (Turbo)12.8B · image · fits from 8 GB
07Short and practical
Guides
G.01What runs on an RTX 5060 Ti 16 GB in ComfyUI: every model, timedMeasuredG.0212 GB or 16 GB VRAM for ComfyUI: what the extra 4 GB gets youBuyingG.03Image editing on 16 GB: FLUX.1 Kontext vs Qwen-Image-Edit 2511, timedMeasuredG.04Best local image model for 16 GB: 9 models tested side by sideMeasuredG.05Can you use it commercially? Licences of local AI modelsLicencesG.06Qwen-Image 2.1, Krea 2 and Ming-Image on 16 GB: timedMeasuredG.07FLUX.2 klein 4B and 9B on a 16 GB RTX 5060 Ti: timedMeasuredG.08Wan 2.2 14B on a 16 GB card: FP8 streaming vs GGUF, timedMeasuredG.09FP8 vs GGUF on a 16 GB RTX 5060 Ti: I timed FLUX.1 devMeasuredG.10GGUF Q8, Q6, Q5, Q4: which one should you download?FilesG.11FP8, GGUF or 16-bit in ComfyUI: what is the difference?FilesG.12How much VRAM do you need for AI images and video in 2026?BuyingG.13Out of memory in ComfyUI: fixes in the order to try themFixesG.14Text encoders: why T5, UMT5 and Qwen-VL eat your memoryExplainerG.15How much system RAM do you need for ComfyUI?BuyingG.16AMD and Intel GPUs for ComfyUI in 2026GPUsG.17Local AI on a laptop: what the GPU name does not tell youGPUsG.1896 GB of "VRAM" on Strix Halo: what fits, and what it costs youGPUsG.19Why video models need so much more memory than image modelsExplainerG.20INT8, NVFP4, MXFP8: the new model file formats of 2026Files