OpenAI: dall-e-3
DALL-E 3 is OpenAI's November 2023 text-to-image model. Official API docs now label it a deprecated image generation model with ID dall-e-3 and say it has been removed from the API. The replacement they name is GPT Image 2. Keep this page as a historical and migration entry, not as a buying guide for new production stills.
Key Features (historical)
- Text-to-image only: Input modality is text; output is image. The model page lists Image generation as supported and Image edit as not supported.
- No Chat Completions or Responses host: Same as later GPT Image models, DALL-E 3 was an Images endpoint model, not a chat model.
- No Batch or fine-tuning: The current model table marks Batch, fine-tuning, and edits as not supported.
- Rate-limit table still published: Docs still show per-tier image RPM, from 1 img/min on the free tier up to 10,000 img/min on Tier 5. That table does not mean you should start a new integration.
Use Cases
- Migration audits: Find leftover
dall-e-3strings in old SDKs and swap them forgpt-image-2. - History and comparison: Explain why 2023-2024 blogs talk about DALL-E 3 when today's default is GPT Image.
- Do not use for new work: Official recommendation is GPT Image 2 for generation and editing.
Pricing
Do not quote a live DALL-E 3 token or per-image price. The model is deprecated and removed. Current OpenAI stills pricing lives on the GPT Image tables (GPT Image 2 standard image output is $30 per 1M tokens as of 2026-08-17). Recheck API pricing when you migrate.
Getting Started (migration)
- Open the DALL-E 3 model page and confirm the deprecated banner.
- Read GPT Image 2 for generate plus edit, sizes up to 4K, and quality presets.
- Replace
model: "dall-e-3"withgpt-image-2and handleb64_jsonthe way the current image guide shows. - Complete Organization Verification if the developer console requires it for GPT Image.
Frequently Asked Questions
Can I still call dall-e-3?
Official docs say the model has been removed from the API. Plan for failure if any old key path still names it.
Is ChatGPT Images 2.0 the same thing?
No. ChatGPT's consumer image UI and the gpt-image-2 API model are related but not the same billable surface.
Why keep this entry?
People still search for DALL-E 3. The useful page now answers "deprecated, use GPT Image 2".
Alternatives
- GPT Image 2: Current OpenAI API stills and edits model.
- Flux.1 Pro: Non-OpenAI commercial stills.
- FLUX 3: If the job has moved from stills into video and audio.
Tips
- Search the repo for
dall-e-3andDALL·Ebefore a release. - Do not promise inpainting on DALL-E 3. Edits were not supported on this model ID.
- Send readers to GPT Image 2 in the first screen, not after a 2023 feature list.
Conclusion
DALL-E 3 mattered in 2023. In 2026 the official line is that it is deprecated and removed. Leave this page up for searchers, then move production stills to GPT Image 2.
Comments
No comments yet. Be the first to comment!
Related Tools
OpenAI: dall-e-2
developers.openai.com/api/docs/models
Previous OpenAI DALL-E 2 snapshot. Rechecked: live image path is GPT Image 2. DALL-E 3 is already removed from the API.
GPT Image 2
developers.openai.com/api/docs/models/gpt-image-2
OpenAI's latest image generation model for fast, high-quality create and edit jobs, with flexible sizes up to 4K and token-based API pricing.
Midjourney V6
www.midjourney.com
Midjourney V6 snapshot from Dec 2023. Rechecked: later versions exist. Not the current default; no invented seat table.
Related Insights
Hook OpenCode Go into Codex on Windows. Do Not Open a Second Toolkit.
A ChatGPT-signed Codex desktop app still shows mostly GPT in the picker. On Windows, enable only OpenCode Go and the Grok, GLM, Kimi, DeepSeek, and MiniMax models you already pay for appear in the same selector. Keys stay local. Native GPT stays put.
What Locks Codex Is the Picker, Not the Models
You already pay for OpenCode Go, Grok, and Z.ai, but the Codex picker still shows mostly GPT. The community project codex-router does not teach another install ritual. It puts subscriptions you already bought back into the selector, keeps keys on the machine, and leaves native GPT alone.
Seven AI Coding CLIs, Six Months: No Matter How Strong the Model, Work Needs Supervision
Claude Code, Codex, opencode, pi, omp and DeepSeek Harness all have personalities. After six months of deep use I run a division of labor: pi for the fastest cheapest reviews, omp for complex PRs, DeepSeek Harness on V4 Flash for high-frequency low-cost review, and Claude Code, Qoder and Cursor for writing. No matter how strong the model, work needs supervision — ideally from an independent third party.