Can the RTX 3060 Ti 8 GB run Krea 2 (Turbo)?
Download Q2_K (4.9 GB). With the model's working memory it needs about 7.5 GB, leaving 0.5 GB spare on 8 GB. Quality: heavy loss; a last resort. For better quality, Q3_K_M (6.0 GB) also runs, with about 0.6 GB spilling into system RAM — a little slower, still practical.
Worth trying on this GPU: with an up-to-date ComfyUI, Dynamic VRAM (on by default for NVIDIA since March 2026) streams whatever does not fit from system RAM, and ComfyUI's own start-up message recommends native FP8/INT8 files over GGUF, saying they “will be faster even if they are larger than your memory”. So before settling for a heavily compressed GGUF, try the native FP8 file (13.1 GB). The file recommended above is the best one that fits entirely — the safe choice on older ComfyUI versions, AMD and Intel. Comfy blog: Dynamic VRAM →
01What to download
| Part | File | Folder | Size | |
|---|---|---|---|---|
| Model | krea2_turbo-Q2_K.gguf Q2_K · vantagewithai/Krea-2-Turbo-GGUF | models/unet | 4.9 GB | Download → |
| Text encoder | qwen3vl_4b_fp8_scaled.safetensors Qwen3-VL 4B FP8 · Comfy-Org/Krea-2 | models/text_encoders | 5.2 GB | Download → |
| VAE | qwen_image_vae.safetensors Qwen-Image VAE · Comfy-Org/Krea-2 | models/vae | 0.3 GB | Download → |
| Total download · keep about the same free on disk | 10.4 GB | |||
Alternatives: Qwen3-VL 4B BF16 (8.9 GB). The 16-bit text encoder is a little more faithful; the smaller one is chosen here because it loads faster and needs less RAM. Sizes read from Hugging Face (2026-09-25). “Download” links start the file directly; the file name links are the same files the official ComfyUI workflows use.
System RAM: 32 GB or more. ComfyUI keeps the model file, the text encoder and the VAE in system RAM and moves them to the GPU as needed. With this set of files that is about 10.4 GB, plus roughly 6 GB for Windows, ComfyUI and a browser: 16.4 GB in total. With less RAM it still runs, but Windows starts swapping to disk and loading gets very slow. calculated
02Every Krea 2 file on 8 GB
| File | Size | Needed | On this card | Quality | Download |
|---|---|---|---|---|---|
| 16-bit SAFETENSORS · Comfy-Org | 26.3 GB | 28.9 GB | 20.9 GB too big | the original weights | Hugging Face → |
| Q8_0 GGUF · vantagewithai | 13.7 GB | 16.3 GB | 8.3 GB too big | practically identical to the original | Hugging Face → |
| FP8 SAFETENSORS · Comfy-Org | 13.1 GB | 15.7 GB | 7.7 GB too big | practically identical to the original | Hugging Face → |
| INT8 SAFETENSORS · Comfy-Org | 13.5 GB | 16.1 GB | 8.1 GB too big | practically identical to the original | Hugging Face → |
| Q6_K GGUF · vantagewithai | 10.6 GB | 13.2 GB | 5.2 GB too big | very close to the original | Hugging Face → |
| Q5_K_M GGUF · vantagewithai | 8.9 GB | 11.5 GB | 3.5 GB too big | close; small differences in fine detail | Hugging Face → |
| Q4_K_M GGUF · vantagewithai | 7.5 GB | 10.1 GB | 2.1 GB too big | good; some loss in fine detail and text | Hugging Face → |
| Q3_K_M GGUF · vantagewithai | 6.0 GB | 8.6 GB | spills 0.6 GB | noticeable loss of detail | Hugging Face → |
| Q2_K ← GGUF · vantagewithai | 4.9 GB | 7.5 GB | fits · 0.5 GB spare | heavy loss; a last resort | Hugging Face → |
Sizes from the Hugging Face file listing, read 2026-09-25. “Needed” = file + 1.8 GB working memory for this model + 0.8 GB kept free for the system.
03The text encoder
Qwen3-VL 4B: 8.9 GB as 16-bit, 5.2 GB as FP8. ComfyUI encodes the prompt first and can push the encoder out of VRAM before sampling, so it does not have to fit together with the model. On 8 GB the FP8 encoder fits on its own, so prompt encoding stays fast.
04About this GPU
The RTX 3060 Ti 8 GB (Ampere) has no FP8 compute. FP8 files still load and save memory, but ComfyUI converts them back before the maths, so there is no speed gain — Q8_0 GGUF is the closer-to-original 8-bit choice. ComfyUI can compute INT8 files natively on NVIDIA, which is worth a try where a model offers one.
RTX 3060 Ti 8 GB: 8 GB GDDR6 · 256-bit · 448 GB/s · Ampere. Everything that runs on the RTX 3060 Ti 8 GB →
05What more VRAM would change
With 16 GB you could run Krea 2 at 8-bit or better (FP8, 13.1 GB) with no offloading: for example on the RTX 5080 16 GB, RTX 5070 Ti 16 GB, RTX 5060 Ti 16 GB.
What changes from the RTX 3060 Ti 8 GB to the RTX 5060 Ti 16 GB →
06Measured and reported results
| Label | Setup | Result | Peak VRAM | Date | Source |
|---|---|---|---|---|---|
| reported | fp8 “3060 ti here, 8GB VRAM, fp8 quant gets about 3.9 seconds per iteration” Posted on Krea-2-Turbo repo; ComfyUI ('run ... in Comfy now'); resolution/steps not stated; year inferred from Krea 2 release (Jun 2026) | 3.9 s/it | — | 2026-07-01 | huggingface.co → |
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.
07Questions
How much VRAM does Krea 2 need?
Around 7.5 GB with the smallest file (Q2_K) and 15.7 GB with an 8-bit file (FP8), counting working memory and a small system reserve. The full 16-bit file needs about 28.9 GB.
Which Krea 2 file should I download for the RTX 3060 Ti 8 GB?
Q2_K (4.9 GB) from vantagewithai/Krea-2-Turbo-GGUF. Load it with Unet Loader (GGUF) from the ComfyUI-GGUF node pack.
Is FP8 faster than GGUF on the RTX 3060 Ti 8 GB?
No. This GPU has no FP8 compute, so ComfyUI converts FP8 weights back before the maths. FP8 only saves memory here; Q8_0 GGUF is the closer-to-original 8-bit pick.
How much do I need to download for Krea 2 on the RTX 3060 Ti 8 GB?
About 10.4 GB for the model file, text encoder and VAE listed on this page, and the same again free on disk. With a 8 GB card, 32 GB of system RAM or more is recommended.