Release Notes

In our ChatGPT output, image_url in ads.advertiser_info has been replaced by image_data, which returns the image itself rather than a link. You still receive everything ChatGPT shows. If your setup reads the old field, switch it to the new one.

You can now connect your AI agents to Web API through a remote MCP server – oxylabs-web-api-mcp, that we host and maintain.

Search and Scrape are available over one MCP connection, with no local deployment or server maintenance on your side. The self-hosted MCP Server remains available for teams that want more control.

You can now see all the web pages Gemini accessed as sources for a response, not just the ones cited inline. Each source comes with a title and URL, so it’s easier to check what shaped the answer. Sources are included whenever Gemini searches the web.

You can now collect video and audio from TikTok and Bilibili with the same Web API workflow you use for other media platforms. Files go straight to your S3 or GCS bucket, ready for AI training, social listening, or trend analysis, with no separate collection or delivery pipelines to build.

You can now get description, badge, url, merchant, offers[], citations[], image_page_title, and image_page_url for every product ChatGPT shows in shopping results. Web API returns it all as structured JSON, so you can see what shapes ChatGPT’s product recommendations without extra setup.

You can now buy and launch Agent Browser straight from the dashboard on your existing Oxylabs subscription, with no separate contract.

When a run fails, you can watch it live or replay it to see exactly what the browser did instead of piecing it together from logs. Persistent profiles let a new session pick up where the previous one left off, instead of starting over each time.