Google has launched an updated version of the Gemini Deep Research agent, competing with OpenAI's GPT-5.2, having extended capabilities for developers.

Google presented on Thursday a revised version of the Gemini Deep Research agent, based on its core model Gemini 3 Pro, on the same day that OpenAI launched GPT-5.2. The new version of Gemini is no longer limited to generating research reports but becomes a developer-ready agent, integrable into third-party applications through the new Interactions API.

Google intends to integrate the agent into its products, such as Google Search and Google Finance. A new evaluation benchmark, DeepSearchQA, has also been introduced, which tests agents on information retrieval tasks. Gemini achieved excellent results, but OpenAI quickly launched GPT-5.2, which claims to surpass competing systems, including Google, on various benchmarks.