OpenAI Deprecates Three Image APIs by December 2026
Key insights
- OpenAI is retiring gpt-image-1-mini, gpt-image-1.5, and chatgpt-image-latest simultaneously on December 1, 2026, all pointing to gpt-image-2.
- DALL-E 2 and DALL-E 3 already retired May 12, 2026; developers who migrated to gpt-image-1-mini now face a second forced move.
- OpenAI's stated policy grants 6+ months advance notice for generally available models, and the June 2 announcement meets that minimum.
Why this matters
OpenAI is systematically narrowing its image generation API to a single model, gpt-image-2, forcing developers to consolidate regardless of which tier they previously built on. Developers who migrated from DALL-E 2 or DALL-E 3 to gpt-image-1-mini as recently as May 2026 now face a second migration within six months, compressing their planning horizon. Combined with the Assistants API discontinuation on August 26, 2026, and a major legacy model sunset on October 23, 2026, the second half of 2026 represents an unusually dense deprecation cycle for anyone building on OpenAI's production APIs.
Summary
OpenAI announced June 2 that three image models shut down December 1, 2026: gpt-image-1-mini, gpt-image-1.5, and chatgpt-image-latest.
All three migrate to gpt-image-2. Some developers face a second forced migration: DALL-E 2 and DALL-E 3 retired May 12, 2026, with gpt-image-1-mini listed as a valid replacement at the time.
Essentially: (OpenAI) is consolidating image generation into a single successor model.
- Three deprecated models, one migration target: gpt-image-2.
- OpenAI provides 6+ months advance notice for generally available models; the June 2 announcement meets that threshold exactly.
- The Assistants API is also discontinued after August 26, 2026, per the same deprecation page.
The pattern is a compressed API surface, retiring legacy and interim models in coordinated waves.
Potential risks and opportunities
Risks
- Developers who built gpt-image-1-mini integrations after the May 2026 DALL-E retirement now face a second migration before December 1, 2026, risking compounded engineering delays.
- Concentrating all image generation traffic on gpt-image-2 creates a single point of failure; if gpt-image-2 has outages or price changes, there is no lower-tier fallback on OpenAI's platform.
- The August 26, 2026 Assistants API discontinuation runs concurrent with the image model timeline, meaning teams may need to migrate multiple systems within the same engineering cycle.
Opportunities
- Third-party API compatibility layer vendors could capture developers who need a migration buffer before the December 1, 2026 gpt-image-1-mini cutoff.
- Alternative image generation API providers could pitch directly to gpt-image-1-mini users who view repeated forced migrations as a reason to diversify away from a single vendor.
- Developer tooling companies offering API version monitoring and dependency scanning gain a clear enterprise pitch given the dense deprecation calendar through late 2026.
What we don't know yet
- Pricing parity: whether gpt-image-2 matches gpt-image-1-mini's cost structure is not disclosed in the deprecation notice.
- Migration tooling: no mention of whether OpenAI will provide automated migration support or compatibility shims before the December 1, 2026 cutoff.
- Downstream impact: how the May 12, 2026 DALL-E retirement affected active developer integrations has not been publicly quantified.
Originally reported by openai.com
Read the original article →Original headline: OpenAI Deprecates gpt-image-1-mini API Model — December 1, 2026 Shutdown, Migration Path to gpt-image-2