{
 "slug": "point-e",
 "name": "Point-E",
 "author": "OpenAI",
 "density": "full",
 "status": "active",
 "status_note": null,
 "alternatives_page": null,
 "one_liner": "Official code and models for generating 3D point clouds from text or images, plus a model that turns point clouds into meshes.",
 "category": {
  "slug": "ai-3d",
  "name": "AI 3D model generators"
 },
 "tags": [
  "Open source"
 ],
 "styles": [],
 "logo_url": "https://agentgid.com/media/avatars/openai.png",
 "logo_letter": "P",
 "links": {
  "repo": "https://github.com/openai/point-e",
  "website": null,
  "api_docs": null
 },
 "preview": null,
 "verdicts": [
  {
   "key": "price",
   "label": "Price",
   "value": "Free",
   "status": "yes",
   "sub": "No paid plan",
   "quote": null,
   "note": "Free open-source code and models under the MIT License.",
   "source": {
    "label": "github.com/openai/point-e",
    "url": "https://github.com/openai/point-e"
   },
   "n": 1
  },
  {
   "key": "commercial",
   "label": "Commercial use",
   "value": "With conditions",
   "status": "cond",
   "sub": "MIT License",
   "quote": "Due to the limitations and biases of our models, we do not currently recommend it for commercial use.",
   "note": "MIT allows it with the notice kept, but OpenAI's model card does not recommend commercial use.",
   "source": {
    "label": "github.com/openai/point-e/blob/main/model-card.md",
    "url": "https://github.com/openai/point-e/blob/main/model-card.md"
   },
   "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/openai/point-e/blob/HEAD/LICENSE"
   },
   "n": 3
  },
  {
   "key": "agent_access",
   "label": "Agent access",
   "value": "Download",
   "status": "info",
   "sub": "Files by hand",
   "quote": null,
   "note": "There is no package or API. Files are downloaded by hand from the official site.",
   "source": {
    "label": "github.com/openai/point-e",
    "url": "https://github.com/openai/point-e"
   },
   "n": 4
  }
 ],
 "license": {
  "name": "MIT License",
  "url": "https://github.com/openai/point-e/blob/HEAD/LICENSE",
  "summary": "The code is under the MIT license, so it may legally be used and changed in any project if the notice stays with it. OpenAI's own model card says it does not recommend the models for commercial use.",
  "pages_read": 2,
  "rules": [
   "Keep the MIT license and copyright notice in all copies or distributions of the code.",
   "Install locally with pip install -e . and run the models yourself; there is no hosted API.",
   "The MIT license covers the software; the pages do not state terms for the generated outputs, so check before relying on them.",
   "OpenAI's model card does not recommend commercial use for now because of quality limits and biases."
  ],
  "rows": [
   {
    "label": "Commercial use",
    "value": "With conditions",
    "status": "cond",
    "quote": "Due to the limitations and biases of our models, we do not currently recommend it for commercial use.",
    "source": {
     "label": "github.com/openai/point-e/blob/main/model-card.md",
     "url": "https://github.com/openai/point-e/blob/main/model-card.md"
    },
    "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/openai/point-e/blob/HEAD/LICENSE"
    },
    "note": null
   },
   {
    "label": "Inside your app or game",
    "value": "Can ship inside your app",
    "status": "yes",
    "quote": "to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies",
    "source": {
     "label": "LICENSE on GitHub",
     "url": "https://github.com/openai/point-e/blob/HEAD/LICENSE"
    },
    "note": null
   },
   {
    "label": "Pass the files on to your users",
    "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/openai/point-e/blob/main/LICENSE"
    },
    "note": "Copying and passing on the code is allowed with the MIT 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": "2026-11-10",
  "commercial_use_page": "https://agentgid.com/assets/point-e/commercial-use/"
 },
 "overview": {
  "heading": "Point-E: rough 3D shapes from text or a picture",
  "body": "Point-E is OpenAI's free research code and models from 2022 that make a rough 3D shape from a short text or a picture. The shape comes out as a cloud of colored dots, and a second small model can turn it into a 3D mesh. The code is MIT licensed, but OpenAI does not recommend it for commercial use."
 },
 "uses": [
  {
   "title": "A 3D shape from a few words",
   "text": "Type a simple description such as \"a red motorcycle\" and get a small colored 3D dot shape of it. It works best with simple objects and colors.",
   "idea": false,
   "image": null
  },
  {
   "title": "A 3D shape from a picture",
   "text": "Give it one rendered picture of an object and get a 3D dot shape that matches it, as a starting point for a model.",
   "idea": false,
   "image": null
  },
  {
   "title": "Turn the dots into a mesh file",
   "text": "Convert the dot shape into a mesh file you can open in other 3D programs, at low or higher detail.",
   "idea": false,
   "image": null
  },
  {
   "title": "Quick prototypes and print sketches",
   "text": "Get a rough first shape of an idea for a prototype or a 3D print sketch, then check and fix it before any real use.",
   "idea": false,
   "image": null
  },
  {
   "title": "A baseline for 3D research",
   "text": "Compare a newer 3D model or your own experiment against a known, open starting point.",
   "idea": false,
   "image": null
  }
 ],
 "start": {
  "prompt": "Clone github.com/openai/point-e, install it with pip install -e . in a virtual environment, run the text2pointcloud notebook steps for the prompt \"a small wooden chair\", then turn the result into mesh.ply with the sdf model at grid_size 128.",
  "install": "pip install -e .",
  "install_note": "Get the code from GitHub and install it from the project folder; there is no ready package to install by name.",
  "code": [
   {
    "tab": "Step 2",
    "intro": "In the text notebook, pick the small text model; its weights download on first use.",
    "source": "base_name = 'base40M-textvec'"
   }
  ]
 },
 "mistakes": [
  {
   "dont": "pip install point-e",
   "fix": "Clone github.com/openai/point-e and run pip install -e . inside it",
   "why": "The project is installed from its source folder; pypi.org had no point-e project in our check on 2026-11-10.",
   "dont_code": true,
   "fix_code": false
  },
  {
   "dont": "Prompting the text model with long, detailed scenes",
   "fix": "Use short prompts naming one simple object and a color",
   "why": "The text model fails on complex prompts and unusual objects.",
   "dont_code": false,
   "fix_code": false
  },
  {
   "dont": "Keeping the example grid_size=32 for a final mesh",
   "fix": "Use grid_size=128 for the detail the paper used",
   "why": "32 is a quick, coarse setting in the example notebook.",
   "dont_code": false,
   "fix_code": false
  },
  {
   "dont": "Running the notebooks from different folders and downloading the models again each time",
   "fix": "Pass one cache_dir to load_checkpoint, or always run from the same folder",
   "why": "By default the model cache is created inside the current working folder.",
   "dont_code": false,
   "fix_code": false
  }
 ],
 "agent": {
  "self_serve": false,
  "self_serve_text": null,
  "install": null,
  "repo": "https://github.com/openai/point-e",
  "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": "3D shapes made with Point-E by OpenAI (MIT License)",
  "rules_text": "- Point-E (AI 3D model generators). Official code and models for generating 3D point clouds from text or images, plus a model that turns point clouds into meshes. Price: free; commercial use with conditions; no credit needed.\n  Connect: download from https://github.com/openai/point-e (no API).\n  Rule: Keep the MIT license and copyright notice in all copies or distributions of the code.\n  Rule: Install locally with pip install -e . and run the models yourself; there is no hosted API.\n  Rule: The MIT license covers the software; the pages do not state terms for the generated outputs, so check before relying on them.\n  Rule: OpenAI's model card does not recommend commercial use for now because of quality limits and biases.\n  License: https://github.com/openai/point-e/blob/HEAD/LICENSE (checked 2026-11-10; details: https://agentgid.com/assets/point-e/)",
  "agent_md": "https://agentgid.com/assets/point-e/agent.md",
  "card_json": "https://agentgid.com/assets/point-e/card.json"
 },
 "popularity": {
  "gid_score": 29,
  "stars": 6903,
  "downloads_7d": null,
  "downloads_7d_label": null,
  "downloads_change_4w_pct": null,
  "eco": null,
  "package": null,
  "repo": "openai/point-e",
  "agent_projects": null,
  "agent_projects_total": null,
  "downloads_unconfirmed": false,
  "series": {
   "downloads_weekly": [],
   "stars": [
    [
     "2026-11-07",
     6902
    ],
    [
     "2026-11-08",
     6902
    ],
    [
     "2026-11-09",
     6902
    ],
    [
     "2026-11-10",
     6902
    ],
    [
     "2026-11-11",
     6904
    ],
    [
     "2026-11-12",
     6905
    ],
    [
     "2026-11-13",
     6903
    ],
    [
     "2026-11-14",
     6903
    ],
    [
     "2026-11-15",
     6901
    ],
    [
     "2026-11-16",
     6901
    ],
    [
     "2026-11-17",
     6903
    ],
    [
     "2026-11-18",
     6903
    ],
    [
     "2026-11-19",
     6903
    ]
   ]
  },
  "stars_flat_since": null,
  "updated_at": "2026-11-19"
 },
 "alternatives": [
  {
   "slug": "point-e",
   "name": "Point-E",
   "why": null,
   "current": true,
   "price": "Free",
   "commercial": {
    "value": "With conditions",
    "status": "cond"
   },
   "access": "Download"
  },
  {
   "slug": "shap-e",
   "name": "Shap-E",
   "why": "Pick Shap-E, OpenAI's later model, when you want textured meshes from text or a picture instead of dot clouds.",
   "current": false,
   "price": "Free",
   "commercial": {
    "value": "With conditions",
    "status": "cond"
   },
   "access": "Download"
  },
  {
   "slug": "triposr",
   "name": "TripoSR",
   "why": "Pick TripoSR when you have a picture and want a mesh file directly, under MIT, on your own computer.",
   "current": false,
   "price": "Free",
   "commercial": {
    "value": "Commercial OK",
    "status": "yes"
   },
   "access": "Download"
  },
  {
   "slug": "trellis",
   "name": "TRELLIS",
   "why": "Pick TRELLIS when you want meshes, 3D Gaussians or radiance fields from a single picture.",
   "current": false,
   "price": "Free",
   "commercial": {
    "value": "With conditions",
    "status": "cond"
   },
   "access": "Download"
  },
  {
   "slug": "hunyuan3d-2",
   "name": "Hunyuan3D 2.0",
   "why": "Tencent's image-to-3D system: a shape model builds the mesh, then a paint model textures it.",
   "current": false,
   "price": "Free",
   "commercial": {
    "value": "With conditions",
    "status": "cond"
   },
   "access": "Download"
  },
  {
   "slug": "depth-anything",
   "name": "Depth Anything 3 (DA3-BASE)",
   "why": "Open-weights vision transformer that predicts depth maps, camera poses and 3D geometry from one or more images.",
   "current": false,
   "price": "Free",
   "commercial": {
    "value": "With conditions",
    "status": "cond"
   },
   "access": "Download"
  }
 ],
 "faq": [
  {
   "q": "Can I use it in a paid product?",
   "a": "The MIT license allows it if you keep the notice, but OpenAI's model card does not recommend commercial use because of the models' limits and biases."
  },
  {
   "q": "Is there an online service or key?",
   "a": "No. You install the code yourself and run the example notebooks; the model files download on first use."
  },
  {
   "q": "What file do I get at the end?",
   "a": "The mesh example writes a PLY file with colors that other 3D programs can import."
  },
  {
   "q": "Which picture model should I use?",
   "a": "The notebook starts with the small base40M model and says base300M or base1B give better results."
  }
 ],
 "at_a_glance": {
  "latest_version": "main branch checked 2026-11-10 (models dated December 2022)",
  "last_commit": "2024-07-04",
  "runtime": "Python with PyTorch, plus OpenAI CLIP installed from GitHub; a CUDA GPU is used when present, otherwise the CPU.",
  "key_needed": false
 },
 "good_to_know": [
  "Results are rough. The shapes have few points and often show noise, stray dots or cracks.",
  "The text model only handles simple descriptions; long or unusual requests fail.",
  "Because it learned from many simple 3D models, results tend to look cartoon-like.",
  "The model files are downloaded from an OpenAI file server into a point_e_model_cache folder inside the folder you run it from.",
  "OpenAI warns against trusting printed parts from it without checking them, and against uses where precision matters."
 ],
 "stack_snippets": [],
 "claims": [
  "The largest picture model has 1 billion parameters",
  "The upsampler takes 1024 points up to 4096 points"
 ],
 "checked": {
  "license_read": "2026-11-10",
  "overview_checked": "2026-11-10",
  "numbers_updated": "2026-11-19",
  "sources": [
   {
    "label": "github.com/openai/point-e",
    "url": "https://github.com/openai/point-e"
   },
   {
    "label": "README on GitHub",
    "url": "https://github.com/openai/point-e/blob/main/README.md"
   },
   {
    "label": "LICENSE on GitHub",
    "url": "https://github.com/openai/point-e/blob/main/LICENSE"
   },
   {
    "label": "github.com/openai/point-e/blob/main/model-card.md",
    "url": "https://github.com/openai/point-e/blob/main/model-card.md"
   },
   {
    "label": "github.com/openai/point-e/blob/main/setup.py",
    "url": "https://github.com/openai/point-e/blob/main/setup.py"
   },
   {
    "label": "github.com/openai/point-e/blob/main/point_e/examples/text2pointcloud.ipynb",
    "url": "https://github.com/openai/point-e/blob/main/point_e/examples/text2pointcloud.ipynb"
   },
   {
    "label": "github.com/openai/point-e/blob/main/point_e/examples/image2pointcloud.ipynb",
    "url": "https://github.com/openai/point-e/blob/main/point_e/examples/image2pointcloud.ipynb"
   },
   {
    "label": "github.com/openai/point-e/blob/main/point_e/examples/pointcloud2mesh.ipynb",
    "url": "https://github.com/openai/point-e/blob/main/point_e/examples/pointcloud2mesh.ipynb"
   },
   {
    "label": "github.com/openai/point-e/blob/main/point_e/models/download.py",
    "url": "https://github.com/openai/point-e/blob/main/point_e/models/download.py"
   },
   {
    "label": "LICENSE on GitHub",
    "url": "https://github.com/openai/point-e/blob/HEAD/LICENSE"
   }
  ],
  "unverified": null
 },
 "user_reports": null,
 "url": "https://agentgid.com/assets/point-e/",
 "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."
}
