Krea 2
Krea 2 (Turbo) needs 13.4 GB of VRAM in ComfyUI, at the diffusion stage of its own template, by file sizes read 3 November 2026. The smallest card that loads it with room to work is the RTX 5080 (16 GB), and the template downloads 18.6 GB.
- Made by
- Krea
- Licence
- Krea 2 Community License
- Released
- June 2026
- Sizes read
- 3 November 2026
Largest stage from exact file sizes, with ComfyUI's own template; LoRAs are not counted. Working memory for the generation comes on top. Open it in the ComfyUI checker
Largest stage from exact file sizes, with ComfyUI's own template; LoRAs are not counted. Working memory for the generation comes on top. Open it in the ComfyUI checker
Which file on your card
For each memory size, the best file for that card: the highest-precision diffusion file that loads with room to work (one that only just fits when nothing else does), and what is left for working memory while it samples. On a larger card that can be above the file ComfyUI's template loads, which the answer above is about. "Streams" means even the smallest file is larger than the card: ComfyUI still runs it from system RAM, more slowly. FP4 files go only to the Blackwell cards that run them natively.
Turbo
ComfyUI's template for Turbo: FP8 scaled diffusion model, FP8 scaled text encoder. Stages: text encoder 5.2 GB, diffusion 13.4 GB. Download 18.6 GB.
Computed 3 November 2026Raw
ComfyUI's template for Raw: BF16 diffusion model, FP8 scaled text encoder. Stages: text encoder 5.2 GB, diffusion 26.5 GB. Download 31.8 GB.
Computed 3 November 2026Check another card, or pick a file by hand, in the ComfyUI VRAM checker. Built your own workflow around it, with LoRAs or custom nodes? Drop it into the ComfyUI workflow checker.
Where to run it
Turbo
On-demand prices before tax, read 19 October to 3 November 2026.
Raw
ComfyUI's Krea 2 (Raw) workflow peaks at 26.5 GB, more than a 24 GB card holds with room to work; a 48 GB card holds it. Every way to run ComfyUI in the cloud, compared.
On-demand prices before tax, read 19 October 2026.
About the model
Krea 2 is Krea's open image model. Krea's open image model from June 2026: a 12-billion-parameter transformer in two versions. Turbo samples in eight steps in Krea's own example and is what ComfyUI's template loads; Raw is the undistilled model, the one trainers point LoRAs at.
Turbo: 12B · Raw: 12B Text encoder: Qwen3-VL 4B Krea 2 Community License Released June 2026
Good at
- Images from 1K up to 2K.
- Style work: Comfy-Org repackages a set of Krea's own LoRAs alongside the model.
- FP8 and INT8 files at about half the size of the BF16 original.
Watch out for
- Commercial use only while your company's revenue over the last twelve months is under US$1 million.
- The NVFP4 and MXFP8 files are made for Blackwell cards (RTX 50, RTX PRO 6000); on anything older, use FP8 or INT8.
- Krea publishes no VRAM figure.
Every Krea 2 file ComfyUI loads
Sizes are the exact byte counts Hugging Face reports, read 3 November 2026. Files marked "template" are the ones ComfyUI's official workflow downloads.
| Part | Precision | Size | Version | File |
|---|---|---|---|---|
| Diffusion model | BF16 | 26.28 GB | Turbo | krea2_turbo_bf16.safetensors |
| Diffusion model | FP8 scaled · template | 13.14 GB | Turbo | krea2_turbo_fp8_scaled.safetensors |
| Diffusion model | INT8 ConvRot | 13.49 GB | Turbo | krea2_turbo_int8_convrot.safetensors |
| Diffusion model | MXFP8 | 13.53 GB | Turbo | krea2_turbo_mxfp8.safetensors |
| Diffusion model | NVFP4 | 7.67 GB | Turbo | krea2_turbo_nvfp4.safetensors |
| Diffusion model | BF16 | 26.28 GB | Raw | krea2_raw_bf16.safetensors |
| Diffusion model | FP8 scaled | 13.14 GB | Raw | krea2_raw_fp8_scaled.safetensors |
| Diffusion model | INT8 ConvRot | 13.49 GB | Raw | krea2_raw_int8_convrot.safetensors |
| Text encoder | BF16 | 8.88 GB | all | qwen3vl_4b_bf16.safetensors |
| Text encoder | FP8 scaled · template | 5.24 GB | all | qwen3vl_4b_fp8_scaled.safetensors |
| VAE | BF16 · template | 0.25 GB | all | qwen_image_vae.safetensors |
What Krea says it needs
Krea states no VRAM figure on either model card or in its GitHub README. The closest it comes: "Architecture: Diffusion Transformer with 12 billion parameters" (source).
Those figures and this page measure different things. A vendor's figure covers its own script at its default resolution, working memory included. This page counts the weights each stage holds, which is exact, and leaves the working memory as the space that remains, because it depends on your resolution and attention kernel and nobody publishes it per model. If generation runs out of memory with a file that loads, lower the resolution before stepping down a precision. ComfyUI's README says it "can run even the biggest open source models on as low as 4GB vram + 8GB ram" by streaming weights (source), so "streams" means slower, not impossible.
LoRA training for Krea 2
Only configurations a trainer's own documentation gives a VRAM figure for. Settings change the figure a great deal, so read the setting column before trusting the number.
| Trainer | VRAM | Setting | Source |
|---|---|---|---|
| diffusion-pipe · Raw | 24 GB | rank 32 at 512 pixels, model held in FP8 | docs |
Questions
- How much VRAM does Krea 2 need?
- In ComfyUI's own workflow for Turbo, the largest stage is the diffusion stage at 13.4 GB, and the whole workflow downloads 18.6 GB. Working memory for generation comes on top and grows with resolution. Krea publishes no VRAM figure.
- Can I run Krea 2 on 8 GB of VRAM?
- Not fully. The smallest file here, FP8 scaled at 13.1 GB, needs 13.4 GB at its largest stage, more than 8 GB holds. ComfyUI will still run it by streaming weights from system RAM, which works but is slower.
- Can I run Krea 2 on 12 GB of VRAM?
- Not fully. The smallest file here, FP8 scaled at 13.1 GB, needs 13.4 GB at its largest stage, more than 12 GB holds. ComfyUI will still run it by streaming weights from system RAM, which works but is slower.
- Can I run Krea 2 on 16 GB of VRAM?
- Yes, with the FP8 scaled file: its largest stage, the diffusion stage, is 13.4 GB, which loads on 16 GB with 2.6 GB left for working memory.
- Which Krea 2 file should I download?
- The highest precision that loads fully on your card. On a 24 GB card that is the FP8 scaled file (13.1 GB). ComfyUI's template downloads FP8 scaled with the FP8 scaled text encoder. FP4 files (NVFP4) are made for Blackwell cards (RTX 50, RTX PRO 6000, DGX Spark); on older cards, use FP8 or INT8.
- Files and sizes: Comfy-Org/Krea-2 on Hugging Face, read from the Hugging Face API. Template files: ComfyUI's Krea 2 guide.
- Model and licence: krea/Krea-2-Turbo and the Krea 2 Community License.
- Card memory: manufacturers' specifications, as on each GPU page. The checker covers NVIDIA cards; ComfyUI also runs on AMD and Apple silicon, where file formats and speed differ and we have not sized them.
- Fit rule: a stage loads fully under 85% of usable memory and is tight up to 95%, the same margins the LLM pages use.
- Prices: each provider's own pricing page, on the dates beside them. Comfy Cloud's answer comes from ComfyUI's own template for that variant, which uses built-in nodes only, and shows only for the variants whose template we checked.