OpenAI adds image results to web search in the Responses API

OpenAI updated its Responses API on June 9, 2026, enabling web search to return image results alongside text. The change is live in the v1/responses endpoint and targets applications where a text-only answer is insufficient — product research, travel, event tracking, and similar use cases where current visual references matter.

What's new

The June 9 changelog entry from OpenAI states: "Web search can now return image results alongside regular text results. Use image search when your application needs current or web-grounded visuals, such as product photos, landmarks, places, events, or visual references."

The feature extends the existing web search tool in the Responses API. Developers can invoke image search for queries where visual results are relevant, receiving image URLs and captions grounded in live web content alongside text citations. Implementation details are available in OpenAI's web search guide.

Context

The Responses API web search capability launched earlier this year to ground model outputs in live web content. The previous implementation returned text excerpts and citations only. Adding image results makes the API output structurally closer to a full web search experience and reduces the need to call a separate image search service.

This is the third developer-facing Responses API update in June alone: moderation scoring for inputs and outputs in a single call (June 4), Lockdown Mode to block prompt injection data exfiltration (June 6), and now image search (June 9). The cadence indicates OpenAI is treating the Responses API as its primary surface for new developer capabilities in 2026.

The timing also coincides with Anthropic's Claude Fable 5 launch and a general acceleration of capability additions across frontier providers. Competitive pressure on feature parity at the API layer is intensifying.

Why it matters

Image search in a single API call opens a practical shortcut for a class of applications that previously required either a separate image search API (Bing Image Search, Google Custom Search, SerpAPI) or multimodal input from pre-fetched screenshots. For developers building AI assistants handling visual queries — shopping, real estate, travel, local discovery — this reduces both the number of API calls and the number of providers needed in a stack.

The change is additive rather than architectural. But it continues a clear pattern: OpenAI is narrowing the gap between what the Responses API can return and what a full browser-grounded agent experience requires, lowering the integration overhead for production applications.