カオスマップに戻る
GPT Image 2.0 (ex-DALL-E)

GPT Image 2.0 (ex-DALL-E)

Overview

DALL-E 3 is OpenAI's text-to-image model that powered ChatGPT image generation and the OpenAI Images API for years. As of 2026 it is effectively sunsetting. In March 2025 ChatGPT switched to GPT-4o native image generation, and in December 2025 DALL-E 3 was removed from ChatGPT entirely. OpenAI announced on November 14, 2025 that DALL-E 3 will be deprecated from the API on May 12, 2026, with developers pushed toward GPT Image 1 / GPT Image 1.5 (gpt-image-1-mini and friends). This page documents DALL-E 3 as a historical reference and migration guide.

Key Features

Strong natural language understanding

DALL-E 3's tight integration with ChatGPT let users refine prompts conversationally. Its prompt-to-image fidelity was class-leading at launch.

Enterprise-friendly safety

The model shipped with relatively strict content filters around people and copyrighted material, making it easier to adopt inside large organisations.

Plans & Pricing (2026 status)

Entry pointPriceNotes
ChatGPT Plus$20/moGPT-Image since Dec 2025
OpenAI API (DALL-E 3)Per-imageSunset May 12, 2026
Successor: gpt-image-1 / 1-miniPer-imageRecommended API route

DALL-E 3 was never sold as a standalone subscription; access was always through ChatGPT or the API.

Typical Use Cases

  • Illustrations for blog posts and documentation
  • Rapid visual prototyping and ideation
  • Casual visuals for decks and social content

Strengths

  • Deep ChatGPT integration enabled conversational iteration
  • Solid prompt comprehension
  • Safer defaults for enterprise environments

Weaknesses

  • API removal on May 12, 2026 — not a safe choice for new projects
  • Already replaced inside ChatGPT by GPT-Image
  • Less distinctive than Midjourney and less photorealistic than Flux

Official Info

公式サイト