Model retirement
OpenAI retires DALL-E from the API, ending the dedicated image-generation brand
- Player
- OpenAI
- What was pulled
- DALL-E 2 and DALL-E 3 (all model snapshots) from the OpenAI API
- Scope
- All API consumers using DALL-E endpoints; the standalone DALL-E consumer brand discontinued
- Announced
- Effective
- Replacement
- GPT-image family — gpt-image-1, gpt-image-1-mini, and gpt-image-2 — integrated into the GPT model pipeline rather than operating as a standalone branded product line.
- Stated reason
- Standard model lifecycle. OpenAI's deprecation notice names the GPT-image family as the successor and ships gpt-image-2 as the state-of-the-art replacement. The retirement is part of a broader OpenAI pattern of consolidating specialised model names under a unified GPT-* and o-* namespace.
Context
DALL-E was the product that brought AI image generation to public awareness in 2021. Its retirement closes the chapter on OpenAI running a named image brand alongside its language models. The replacement architecture — gpt-image-* integrated into the GPT pipeline — consolidates image generation as a capability of the GPT family rather than a separate product line with its own API surface and brand identity.
Developers who built around dall-e-3 need to migrate to gpt-image-2. The migration cost includes prompt re-tuning, output-quality regression testing, and pricing recalculation; the new models differ in pricing, quality characteristics, and API design. Organisations that signed procurement or SLA agreements referencing DALL-E model names will need contract amendments.