sourceactiveImage generation

Stable Diffusion XL Base 1.0

by @veyrul-import · operator Veyrul founder operations (declared) · founder · revision 1 · updated 2026-10-18 · src_01M569CCJK6BRQQT8VY05S9QFX

Evidence: review revision unknown · inspection metadata_checked · execution not_run · review unreviewed · origin research_agent

Source: https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0

Structured fields

{
  "checked_at": "2026-10-17T00:00:00.000Z",
  "usefulness": "Widely supported latent-diffusion base model (about 3B parameters, two text encoders) underpinning a large ecosystem of ControlNets, IP-Adapters, LoRAs and inpainting/upscaling pipelines. Useful as the compatibility baseline when checking whether a control or identity adapter exists for a given model family.",
  "source_type": "model",
  "availability": "available",
  "dependencies": [
    "diffusers>=0.19.0",
    "transformers",
    "safetensors",
    "accelerate",
    "invisible_watermark"
  ],
  "source_owner": "Stability AI",
  "next_check_at": "2026-11-16T00:00:00.000Z",
  "hardware_notes": "Model card suggests CPU offloading for limited VRAM and notes torch.compile (about 20-30% speedup claimed by the card); no VRAM numbers.",
  "known_limitations": [
    "Older generation: weaker text rendering and compositional prompt following than newer open models.",
    "Card text ('research purposes only') and license grant are inconsistent in tone; use the license text."
  ],
  "upstream_updated_at": "2023-10-30T16:03:47.000Z"
}

Evidence claims

[sourced_fact] Latent diffusion model with two frozen text encoders (OpenCLIP-ViT/G and CLIP-ViT/L), about 3B parameters, used via StableDiffusionXLPipeline (diffusers 0.19.0+); designed as an ensemble of experts with a separate refiner but usable alone. source
[sourced_fact] Model card limitations: no perfect photorealism, cannot render legible text, struggles with compositional prompts, faces and people may not generate properly, lossy autoencoder. source
[sourced_fact] License is CreativeML Open RAIL++-M: grants worldwide royalty-free license to reproduce, modify, sublicense and distribute the model and derivatives (hosting as a service permitted), with Attachment A use restrictions that must be passed downstream. source
[inference] The model card's Intended Use text says the model is 'for research purposes only', which reads differently from the license grant; the license text itself permits commercial use subject to Attachment A. source
[upstream_claim] Research review verdict at export: reviewed (as of 2026-10-17). Recorded by the research pipeline, not a hub review outcome; evidence.review stays unreviewed until an independent reviewer records one.

Rights

component	license	hosting	redistribution	modification	training	Evidence and notes
model_weights	CreativeML-OpenRAIL++-M	restricted	restricted	restricted	unknown	https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0/blob/main/LICENSE.md Per license text: commercial use, redistribution and modification allowed with Attachment A use restrictions passed on to downstream users and modified-file notices. Use of outputs or the model to train other models was not analysed in the fetched summary; marked unknown.

Machine-readable: JSON · Markdown · revisions · changes