# DALL-E Launch DALL-E is a series of text-to-image models developed by OpenAI using deep learning to generate digital images from natural language prompts. The first version was announced in January 2021, with successors DALL-E 2 and DALL-E 3 released in 2022 and 2023 respectively. DALL-E, DALL-E 2, and DALL-E 3 (stylised DALL·E) are text-to-image models developed by [OpenAI](https://www.wikiprompt.org/wiki/openai) using [deep learning](https://www.wikiprompt.org/wiki/deep-learning) methodologies to generate digital images from natural language descriptions known as prompts. The first version was announced in January 2021, and subsequent iterations expanded capabilities in realism, resolution, and prompt following. The name is a portmanteau of the Pixar robot character WALL-E and the Spanish surrealist artist Salvador Dalí. ## History and background OpenAI revealed DALL-E in a blog post on 5 January 2021, using a modified version of [GPT-3](https://www.wikiprompt.org/wiki/transformer) to generate images. On 6 April 2022, the company announced DALL-E 2, a successor designed to produce more realistic images at higher resolutions that could "combine concepts, attributes, and styles." Access to DALL-E 2 initially remained restricted to pre-selected users for a research preview due to ethical and safety concerns. On 20 July 2022, the model entered a beta phase with invitations sent to 1 million waitlisted individuals, allowing a certain number of free generations monthly with options to purchase more. The waitlist requirement was removed on 28 September 2022, opening the tool to everyone. In September 2023, OpenAI announced DALL-E 3, capable of understanding "significantly more nuance and detail" than previous iterations. It was released natively into ChatGPT for Plus and Enterprise customers in October 2023, with availability via the API and a "Labs" platform in early November. Microsoft implemented the model in Bing's Image Creator tool and its Designer app, with Microsoft Copilot running on DALL-E 3. In March 2025, DALL-E 3 was replaced in ChatGPT by GPT Image's native image-generation capabilities. In February 2024, OpenAI began adding watermarks to DALL-E generated images, containing metadata in the C2PA (Coalition for Content Provenance and Authenticity) standard. ## Technology The first generative pre-trained transformer (GPT) model was developed by OpenAI in 2018 using a [Transformer architecture](https://www.wikiprompt.org/wiki/transformer). Scaling iterations produced GPT-2 in 2019 and GPT-3 in 2020, the latter with 175 billion parameters. DALL-E built on this foundation with distinct architectures across versions. ### DALL-E The original DALL-E comprised three components: a discrete variational autoencoder (VAE), an autoregressive decoder-only Transformer model with 12 billion parameters similar to GPT-3, and a CLIP pair of image and text encoders. The discrete VAE converted images to sequences of tokens and back, necessary because the Transformer does not directly process image data. Input consisted of a tokenised English caption (up to 256 tokens, vocabulary size 16,384) followed by tokenised image patches. Each 256×256 RGB image was divided into 32×32 patches of 4×4 pixels, with each patch converted to a token using a vocabulary of 8,192. CLIP (Contrastive Language-Image Pre-training), developed and announced alongside DALL-E, was a separate model based on [contrastive learning](https://www.wikiprompt.org/wiki/contrastive-learning) trained on 400 million image-text pairs scraped from the Internet. It ranked DALL-E's output by predicting which caption from a list of 32,768 randomly selected captions was most appropriate for an image, filtering a larger initial list to select the closest match to the prompt. ### DALL-E 2 DALL-E 2 used 3.5 billion parameters, fewer than its predecessor, and replaced the autoregressive Transformer with a [diffusion model](https://www.wikiprompt.org/wiki/diffusion-model) conditioned on CLIP image embeddings. During inference, these embeddings were generated from CLIP text embeddings by a prior model. This architecture was the same as that of Stable Diffusion, released a few months later. ### DALL-E 3 OpenAI published a technical report for DALL-E 3, but it omitted training and implementation details, focusing instead on improved prompt-following capabilities. The model was integrated into ChatGPT Plus, enabling conversational refinement of generated images. ## Capabilities DALL-E could generate imagery in multiple styles, including photorealistic images, paintings, and emoji. It could "manipulate and rearrange" objects and place design elements in novel compositions without explicit instruction. Thom Dunn, writing for BoingBoing, noted that when asked to draw a daikon radish blowing its nose, sipping a latte, or riding a unicycle, DALL-E often drew the handkerchief, hands, and feet in plausible locations. The model could "fill in the blanks," inferring appropriate details such as Christmas imagery for related prompts or shadows not mentioned in the text, and exhibited broad understanding of visual and design trends. DALL-E could produce images for a wide variety of arbitrary descriptions from various viewpoints with only rare failures. Mark Riedl, an associate professor at the Georgia Tech School of Interactive Computing, found that DALL-E could blend concepts, a key element of human creativity. Its visual reasoning ability was sufficient to solve Raven's Matrices, visual tests often administered to measure human intelligence. DALL-E 3 followed complex prompts with more accuracy and detail than predecessors and generated more coherent and accurate text within images. ## Image modification Given an existing image, DALL-E 2 and DALL-E 3 could produce "variations" as individual outputs based on the original, as well as edit the image to modify or expand it. The "inpainting" and "outpainting" abilities used context from the image to fill missing areas in a medium consistent with the original, following a given prompt. For example, this allowed inserting a new subject into an image or expanding beyond its original borders. According to OpenAI, "Outpainting takes into account the image's existing visual elements - including shadows, reflections, and textures - to" produce seamless extensions. --- Source: https://www.wikiprompt.org/wiki/dall-e-launch-2021 License: CC BY-SA 4.0 (https://creativecommons.org/licenses/by-sa/4.0/) Last updated: 2026-09-09T02:01:43.180541+00:00