Google Builds Gemini API Docs Directly Into AI Studio

Google just shipped a preview that removes one of the more annoying context switches in Gemini development: bouncing between AI Studio and a separate documentation tab. The full Gemini API reference now lives inside AI Studio, sitting alongside the prompt editor, model picker, and code view where you actually build.

Rather than treating documentation as a destination you visit, the team is treating it as a panel you keep open while you work, so reference material stays adjacent to whatever you are prototyping.

What actually changed

Until now, AI Studio has been the fast-iteration surface for Gemini, while ai.google.dev hosted the reference material. That split made sense when AI Studio was mostly a prompt playground, but the product has grown considerably. It now covers Gemini 3.1 Pro previews, Veo video models, Nano Banana image generation, embeddings, robotics endpoints, and the Interactions API. With Google positioning the Gemini API as the fastest path from prompt to production across text, image, and multimodal workloads, hunting through a separate site for the right function-calling flag or billing tier detail became friction. Merging the two views fixes that.

Why the timing fits

The Gemini API has been shipping features at an aggressive cadence, and many of them are configuration choices you need documentation open to get right. Recent additions include:

• Built-in tools combined with function calling, so Gemini's native tools can run alongside custom function-calling tools in a single API call
• Flex and Priority inference tiers, giving you more control over cost or latency
• gemini-embedding-2-preview, a multimodal embedding model that maps text, image, video, audio, and PDF inputs into a unified embedding space
• Grounding with Google Maps for Gemini 3 models going forward
• Project-level spend caps in AI Studio billing

Each of these has knobs, model IDs, and quota rules that are easy to misconfigure. Reading the reference in the same window where you are editing a prompt or inspecting a request payload removes a class of copy-paste errors.

Who should care

If you already live in AI Studio for prototyping, this is a pure quality-of-life upgrade. It helps most when you are:

1. Wiring up function calling or tool use and need to check the exact schema shape.
2. Picking between preview and stable model aliases, where previews usually have billing enabled, ship with tighter rate limits, and get deprecated with at least two weeks notice.
3. Debugging a request and cross-checking parameters without losing your current session.
4. Onboarding to a capability like multimodal embeddings or robotics endpoints where you have no muscle memory yet.

The limits of a preview

This is labeled as a first preview, meaning the in-Studio documentation is effectively a mirror of the public site with a new host surface, not a rewrite. Expect it to change where you read the reference, not what it says. If your workflow already leans on browser tabs, bookmarks, or search across ai.google.dev, the win is smaller. And because AI Studio itself remains free for experimentation within usage limits, embedding the reference inside it does not change pricing or quotas for the API you eventually call.

For a team that runs its whole Gemini prototyping loop inside AI Studio, having the reference one click away rather than one tab away is the kind of small change that compounds over a week of building.