{
 "slug": "if",
 "name": "DeepFloyd IF",
 "author": "DeepFloyd Lab at StabilityAI",
 "density": "full",
 "status": "active",
 "status_note": null,
 "alternatives_page": null,
 "one_liner": "Open-source text-to-image diffusion model with a T5 text encoder and three cascaded stages, producing up to 1024x1024 px images.",
 "category": {
  "slug": "ai-images",
  "name": "AI image generators"
 },
 "tags": [
  "Code library",
  "Open source"
 ],
 "styles": [],
 "logo_url": "https://agentgid.com/media/avatars/deep-floyd.png",
 "logo_letter": "D",
 "links": {
  "repo": "https://github.com/deep-floyd/IF",
  "website": null,
  "api_docs": null
 },
 "preview": null,
 "verdicts": [
  {
   "key": "price",
   "label": "Price",
   "value": "Free",
   "status": "yes",
   "sub": "No paid plan",
   "quote": null,
   "note": null,
   "source": {
    "label": "github.com/deep-floyd/IF",
    "url": "https://github.com/deep-floyd/IF"
   },
   "n": 1
  },
  {
   "key": "commercial",
   "label": "Commercial use",
   "value": "With conditions",
   "status": "cond",
   "sub": "Modified MIT",
   "quote": "to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software",
   "note": "The model and the data it produces may not be used for any commercial or production purposes.",
   "source": {
    "label": "LICENSE on GitHub",
    "url": "https://github.com/deep-floyd/IF/blob/HEAD/LICENSE"
   },
   "n": 2
  },
  {
   "key": "credit",
   "label": "Credit the author",
   "value": "Not needed",
   "status": "yes",
   "sub": "Keep the notice in code copies",
   "quote": "The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.",
   "note": "No credit on screen. Keep the license notice inside the code you ship.",
   "source": {
    "label": "LICENSE on GitHub",
    "url": "https://github.com/deep-floyd/IF/blob/HEAD/LICENSE"
   },
   "n": 3
  },
  {
   "key": "agent_access",
   "label": "Agent access",
   "value": "PyPI package",
   "status": "info",
   "sub": "Installs without a key",
   "quote": null,
   "note": "An agent can install it by itself with pip install deepfloyd_if.",
   "source": {
    "label": "github.com/deep-floyd/IF",
    "url": "https://github.com/deep-floyd/IF"
   },
   "n": 4
  }
 ],
 "license": {
  "name": "Modified MIT (code); separate DeepFloyd IF license for weights",
  "url": "https://github.com/deep-floyd/IF/blob/HEAD/LICENSE",
  "summary": "The trained model is for non-commercial research only, and even its pictures may not be used for commercial purposes. The code has a modified MIT license that forbids turning off its safety filters.",
  "pages_read": 1,
  "rules": [
   "Keep the copyright and license notice in all copies of the code you ship.",
   "Do not remove, disable or circumvent the inference filters (safety checker, watermarker) in the Software.",
   "Accept the DeepFloyd IF weights license on Hugging Face before downloading the model weights, and check LICENSE-MODEL for their terms."
  ],
  "rows": [
   {
    "label": "Commercial use",
    "value": "With conditions",
    "status": "cond",
    "quote": "to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software",
    "source": {
     "label": "LICENSE on GitHub",
     "url": "https://github.com/deep-floyd/IF/blob/HEAD/LICENSE"
    },
    "note": null
   },
   {
    "label": "Credit",
    "value": "Not needed",
    "status": "yes",
    "quote": "The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.",
    "source": {
     "label": "LICENSE on GitHub",
     "url": "https://github.com/deep-floyd/IF/blob/HEAD/LICENSE"
    },
    "note": null
   },
   {
    "label": "Inside your app or game",
    "value": "With conditions",
    "status": "cond",
    "quote": "to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software",
    "source": {
     "label": "LICENSE on GitHub",
     "url": "https://github.com/deep-floyd/IF/blob/HEAD/LICENSE"
    },
    "note": null
   },
   {
    "label": "Pass the files on to your users",
    "value": "With conditions",
    "status": "cond",
    "quote": "Together with any copies of the Software Products (as well as derivative works thereof or works incorporating the Software Products) that you distribute, you must provide (i) a copy of this License,",
    "source": {
     "label": "github.com/deep-floyd/IF/blob/develop/LICENSE-MODEL",
     "url": "https://github.com/deep-floyd/IF/blob/develop/LICENSE-MODEL"
    },
    "note": "Non-commercial research copies only, with a copy of the license and the DeepFloyd attribution notice."
   },
   {
    "label": "AI training",
    "value": "Not stated",
    "status": "unknown",
    "quote": null,
    "source": null,
    "note": null
   },
   {
    "label": "Bots and bulk downloads",
    "value": "Not stated",
    "status": "unknown",
    "quote": null,
    "source": null,
    "note": null
   }
  ],
  "checked_at": "2026-10-24",
  "commercial_use_page": "https://agentgid.com/assets/if/commercial-use/"
 },
 "overview": {
  "heading": "DeepFloyd IF: research text-to-image model, non-commercial",
  "body": "DeepFloyd IF is an AI model from DeepFloyd Lab and Stability AI that draws a picture from a text description, up to 1024 by 1024 pixels. It runs on your own computer with a large NVIDIA graphics card. The code is open, but the trained model may be used only for non-commercial research."
 },
 "uses": [
  {
   "title": "Pictures from a text description",
   "text": "Write what you want to see, and the model draws it in three steps, from a small picture up to 1024 by 1024 pixels. Research use only.",
   "idea": false,
   "image": null
  },
  {
   "title": "Restyle an existing picture",
   "text": "Give it a picture and a style such as \"origami\" or \"oil painting\", and it redraws the picture in that style.",
   "idea": false,
   "image": null
  },
  {
   "title": "Enlarge a small picture",
   "text": "Make a small picture four times larger with added detail, even if the model did not draw it.",
   "idea": false,
   "image": null
  },
  {
   "title": "Repaint one part of a picture",
   "text": "Mark an area of a picture and describe what should be there, and the model fills only that area.",
   "idea": false,
   "image": null
  },
  {
   "title": "Teach it a new subject",
   "text": "Train it on a few photos of one object or character so it can draw that subject in new scenes, for research.",
   "idea": false,
   "image": null
  },
  {
   "title": "Compare image models in a study",
   "text": "Use it as a fixed, well-known model next to newer ones when you compare how models follow prompts.",
   "idea": true,
   "image": null
  }
 ],
 "start": {
  "prompt": "For a non-commercial research test only, set up DeepFloyd IF with Diffusers in a Python virtual environment, read my Hugging Face token from the HF_TOKEN environment variable, and run stage 1 and stage 2 from the official README on my prompt.",
  "install": null,
  "install_note": null,
  "code": [
   {
    "tab": "Step 2",
    "intro": "Install the Diffusers library and its helpers in a Python environment with PyTorch for your graphics card.",
    "source": "pip install diffusers accelerate transformers safetensors"
   },
   {
    "tab": "Step 3",
    "intro": "Log in from Python with your own access token, then run the three stages from the README example.",
    "source": "from huggingface_hub import login\n\nlogin()"
   }
  ]
 },
 "mistakes": [
  {
   "dont": "Using IF pictures in a client's ad, product page or paid app",
   "fix": "Use it only for non-commercial research; pick a model whose license allows commercial use for production",
   "why": "The model license bans commercial and production use of the model and of data it produces.",
   "dont_code": false,
   "fix_code": false
  },
  {
   "dont": "Calling from_pretrained(\"DeepFloyd/IF-I-XL-v1.0\") before the person has accepted the license and logged in",
   "fix": "Have the person accept the license on the model page, then log in with huggingface_hub login() or an HF_TOKEN",
   "why": "The weights are gated; the download fails without an accepted license and a token.",
   "dont_code": false,
   "fix_code": false
  },
  {
   "dont": "Keeping enable_xformers_memory_efficient_attention() on torch 2.0 or newer",
   "fix": "Delete those lines when torch is 2.0.0 or newer",
   "why": "The README says to delete all of them on newer torch.",
   "dont_code": false,
   "fix_code": false
  },
  {
   "dont": "Passing safety_checker=None or removing the watermarker to speed things up",
   "fix": "Keep the safety checker and watermarker, as in the README's stage 3 example",
   "why": "The code license forbids disabling the inference filters.",
   "dont_code": false,
   "fix_code": false
  }
 ],
 "agent": {
  "self_serve": true,
  "self_serve_text": "Agent installs a package or loads it from a CDN",
  "install": "pip install deepfloyd_if",
  "repo": "https://github.com/deep-floyd/IF",
  "api_docs": null,
  "api_account": "Not needed. Everything runs on your side.",
  "get_files": null,
  "formats": [],
  "limits": null,
  "mcp": null,
  "mcp_url": null,
  "credit_line": "DeepFloyd is licensed under the DeepFloyd License, Copyright (c) Stability AI Ltd. All Rights Reserved.",
  "rules_text": "- DeepFloyd IF (AI image generators). Open-source text-to-image diffusion model with a T5 text encoder and three cascaded stages, producing up to 1024x1024 px images. Price: free; commercial use with conditions; no credit needed; ship inside your app with conditions.\n  Connect: PyPI deepfloyd_if.\n  Rule: Keep the copyright and license notice in all copies of the code you ship.\n  Rule: Do not remove, disable or circumvent the inference filters (safety checker, watermarker) in the Software.\n  Rule: Accept the DeepFloyd IF weights license on Hugging Face before downloading the model weights, and check LICENSE-MODEL for their terms.\n  License: https://github.com/deep-floyd/IF/blob/HEAD/LICENSE (checked 2026-10-24; details: https://agentgid.com/assets/if/)",
  "agent_md": "https://agentgid.com/assets/if/agent.md",
  "card_json": "https://agentgid.com/assets/if/card.json"
 },
 "popularity": {
  "gid_score": 16,
  "stars": 7779,
  "downloads_7d": 4,
  "downloads_7d_label": "4",
  "downloads_change_4w_pct": null,
  "eco": "pypi",
  "package": "deepfloyd_if",
  "repo": "deep-floyd/IF",
  "agent_projects": null,
  "agent_projects_total": null,
  "downloads_unconfirmed": false,
  "series": {
   "downloads_weekly": [
    [
     "2026-04-20",
     31
    ],
    [
     "2026-06-15",
     65
    ]
   ],
   "stars": [
    [
     "2026-11-07",
     7784
    ],
    [
     "2026-11-08",
     7784
    ],
    [
     "2026-11-09",
     7784
    ],
    [
     "2026-11-10",
     7781
    ],
    [
     "2026-11-11",
     7782
    ],
    [
     "2026-11-12",
     7782
    ],
    [
     "2026-11-13",
     7781
    ],
    [
     "2026-11-14",
     7781
    ],
    [
     "2026-11-15",
     7780
    ],
    [
     "2026-11-16",
     7780
    ],
    [
     "2026-11-17",
     7779
    ],
    [
     "2026-11-18",
     7779
    ],
    [
     "2026-11-19",
     7779
    ]
   ]
  },
  "stars_flat_since": null,
  "updated_at": "2026-11-19"
 },
 "alternatives": [
  {
   "slug": "if",
   "name": "DeepFloyd IF",
   "why": null,
   "current": true,
   "price": "Free",
   "commercial": {
    "value": "With conditions",
    "status": "cond"
   },
   "access": "PyPI"
  },
  {
   "slug": "sdxl",
   "name": "Stable Diffusion XL base 1.0",
   "why": "Pick SDXL when you need an open image model whose weights allow commercial use under its own license.",
   "current": false,
   "price": "Free",
   "commercial": {
    "value": "With conditions",
    "status": "cond"
   },
   "access": "Download"
  },
  {
   "slug": "flux-1-dev",
   "name": "FLUX.1 [dev]",
   "why": "Pick FLUX.1 [dev] for a newer open-weights model whose license is also non-commercial but lets you use the pictures commercially with some limits.",
   "current": false,
   "price": "Free",
   "commercial": {
    "value": "Non-commercial",
    "status": "no"
   },
   "access": "Download"
  },
  {
   "slug": "openai-gpt-image",
   "name": "OpenAI GPT Image",
   "why": "Pick OpenAI's image model when you want a paid hosted service with no graphics card of your own.",
   "current": false,
   "price": "Paid",
   "commercial": {
    "value": "Commercial OK",
    "status": "yes"
   },
   "access": "API, npm, PyPI"
  },
  {
   "slug": "invokeai",
   "name": "InvokeAI",
   "why": "Free open-source, self-hosted AI image generation app with a layer-based canvas and node-based workflows.",
   "current": false,
   "price": "Free",
   "commercial": {
    "value": "Commercial OK",
    "status": "yes"
   },
   "access": "Download"
  },
  {
   "slug": "z-image-turbo",
   "name": "Z-Image-Turbo",
   "why": "Efficient open-weights image generation model built on a single-stream diffusion transformer.",
   "current": false,
   "price": "Free",
   "commercial": {
    "value": "Commercial OK",
    "status": "yes"
   },
   "access": "Download"
  }
 ],
 "faq": [
  {
   "q": "Can I use the pictures in ads, a shop or a paid app?",
   "a": "No. The model license forbids commercial or production use, and this also covers data the model produces."
  },
  {
   "q": "How big are the pictures?",
   "a": "Stage 1 draws 64 by 64 pixels, and two more stages enlarge it to 256 and then 1024 pixels."
  },
  {
   "q": "Do I need an account?",
   "a": "Yes. You need a Hugging Face account, accept the license on the model page and log in locally with a token."
  },
  {
   "q": "What graphics card do I need?",
   "a": "16 GB of graphics memory for the first two stages and 24 GB with the 1024-pixel stage; Diffusers can run it with about 14 GB."
  },
  {
   "q": "Must I credit DeepFloyd when I share it?",
   "a": "Yes. When you pass on the model or works that include it, add a copy of the license and the attribution notice it gives."
  }
 ],
 "at_a_glance": {
  "latest_version": "deepfloyd_if 1.0.2rc0 as the README installs it (latest stable on PyPI 1.0.1); weights revision c03d510e9b75bce9f9db5bb85148c1402ad7e694",
  "last_release": "2023-04-28",
  "last_commit": "2024-04-14",
  "runtime": "deepfloyd_if 1.0.1 needs torch below 2.0.0 and diffusers ~0.16.0; with diffusers on torch 2.0 or newer, drop the xformers lines. NVIDIA GPU with 14-24 GB VRAM.",
  "key_needed": false
 },
 "good_to_know": [
  "Not for paid products. The model license bans commercial and production use of the model and of the pictures it makes.",
  "You need a Hugging Face account, must accept the license on the model page, and must log in with an access token before the model downloads.",
  "It needs a large graphics card. The full set of models needs 24 GB of graphics memory; Diffusers can run it with about 14 GB.",
  "The safety checker and the watermark must stay on. The code license forbids removing or turning off these filters.",
  "The project looks finished rather than active. The model files date from June 2023 and the GitHub code was last changed in April 2024."
 ],
 "stack_snippets": [],
 "claims": [
  "Zero-shot FID score of 6.66 on the COCO dataset",
  "The largest first-stage model, IF-I-XL, has 4.3B parameters"
 ],
 "checked": {
  "license_read": "2026-10-24",
  "overview_checked": "2026-11-12",
  "numbers_updated": "2026-11-19",
  "sources": [
   {
    "label": "github.com/deep-floyd/IF",
    "url": "https://github.com/deep-floyd/IF"
   },
   {
    "label": "README on GitHub",
    "url": "https://github.com/deep-floyd/IF/blob/develop/README.md"
   },
   {
    "label": "LICENSE on GitHub",
    "url": "https://github.com/deep-floyd/IF/blob/develop/LICENSE"
   },
   {
    "label": "github.com/deep-floyd/IF/blob/develop/LICENSE-MODEL",
    "url": "https://github.com/deep-floyd/IF/blob/develop/LICENSE-MODEL"
   },
   {
    "label": "api.github.com/repos/deep-floyd/IF",
    "url": "https://api.github.com/repos/deep-floyd/IF"
   },
   {
    "label": "pypi.org/pypi/deepfloyd-if/json",
    "url": "https://pypi.org/pypi/deepfloyd-if/json"
   },
   {
    "label": "huggingface.co/DeepFloyd/IF-I-XL-v1.0",
    "url": "https://huggingface.co/DeepFloyd/IF-I-XL-v1.0"
   },
   {
    "label": "huggingface.co",
    "url": "https://huggingface.co/api/models/DeepFloyd/IF-I-XL-v1.0"
   },
   {
    "label": "huggingface.co",
    "url": "https://huggingface.co/docs/diffusers/api/pipelines/deepfloyd_if"
   },
   {
    "label": "LICENSE on GitHub",
    "url": "https://github.com/deep-floyd/IF/blob/HEAD/LICENSE"
   }
  ],
  "unverified": null
 },
 "user_reports": null,
 "url": "https://agentgid.com/assets/if/",
 "note": "Written by AgentGid from the official pages in checked.sources; quotes are copied from those pages. status: yes | cond | no | unknown | info. Prices and vendor claims are as reported on the checked date. Not legal advice."
}
