Vercel Connect
Manage runtime credentials and third-party service connections for applications and AI agents across projects and environments.
Gemini API
ai.google.dev
Google’s developer platform provides models and APIs for multimodal applications, generated media and agents.
Visit website
Pricing
Freemium
Available on
API, Google AI Studio
Category
Developer Tools
What is Gemini API?
Gemini API is Google’s developer platform for building AI applications and agents. Its portfolio covers coding, multimodal reasoning, live voice and video, and image, video and music generation through different models and APIs.
Developers can use Google AI Studio to test prompts, manage API keys, monitor usage and build prototypes. Google recommends the Interactions API for new projects; generateContent remains supported as a legacy interface.
Key features
Multimodal applications
Build applications around text and other media using a model and interface that support the required input and output types.
Tools and structured output
The getting-started guide covers structured responses, tools and function calling for connecting model responses to application workflows.
Agents and background execution
Developers can explore managed agents and background execution for tasks that extend beyond a single immediate response.
Live and generated media
The platform includes a Live API for voice and video, alongside separate image, video and music generation offerings.
Logs and datasets
Paid-tier projects can inspect request logs, configure storage and save logs into datasets for later analysis.
Who is it for?
Application features
A developer could add structured content extraction or multimodal question answering to an existing application.
Media workflows
A creative tools team could prototype image or other media generation with an appropriate model.
Agent workflows
An engineering team could explore a tool-connected agent for a defined multistep task.
How to use Gemini API
- Open Google AI Studio and create or select a project and API key.
- Choose a model and confirm that it supports your required capabilities and pricing tier.
- Use an SDK such as Python or JavaScript, or REST, to send an initial request through the Interactions API.
- Review the response and usage, then add structured output or tools as needed.
- Check project logging settings and request-level storage controls before integrating production traffic.
Gemini API pricing
Freemium
The Free tier provides limited model access. Paid production usage offers prepaid and pay-as-you-go billing with higher limits. Prices depend on the selected model and service; free-tier content is used for product improvement, while paid-tier content is not.
Gemini 4 Argon Standard paid-tier usage
USD 2 per million input tokens and USD 10 per million output tokens
Usage-based token billing
Rates apply through November 30, 2026. Output billing includes thinking tokens. From December 1, 2026, the listed rates become USD 4 per million input tokens and USD 20 per million output tokens. Argon has no Free tier.
What to know before you use it
- Capabilities, prices and availability vary across models and services. Free-tier access is limited to certain models.
- Gemini API and Google AI Studio are available only in supported countries and territories, with additional access requirements.
- Developer-visible log storage requires a paid-tier project. Interactions requests are stored by default. Generate Content also stores requests by default for projects created on or after October 26, 2026; older projects retain their existing setting. Project controls and request-level overrides are available.
- Logging excludes some model families, public-preview agents and inputs containing videos, GIFs or PDFs. These exclusions concern logging rather than general API input support.
Frequently asked questions
Which interface should a new project use?
Google recommends the Interactions API. Existing generateContent integrations remain supported, with migration guidance available.
Can default request storage be disabled?
Yes. Interactions stores requests by default; Generate Content does so for projects created on or after October 26, 2026, while older projects retain their setting. Project controls manage each API independently, and a per-request store setting overrides them. Generate Content request-level controls require a supported SDK version; consult the official Logs and datasets guide for version requirements. Disabling Interactions storage prevents automatic conversation-history storage and retrieval unless overridden per request.
How long are logs retained?
Logs expire after 55 days by default, with configurable retention. Logs saved into datasets do not expire and can be exported as CSV, JSONL or Google Sheets.