Skip to content

SaaS Ecosystem & Business · Freemium

Gemini API

The Gemini API is Google's developer API for its Gemini models, used through Google AI Studio and the Google Gen AI SDKs. Developers use it to generate and reason over text, images, audio, video and documents, call functions and built-in tools such as Grounding with Google Search, create embeddings, and run batch jobs. It is distinct from the Gemini consumer app, and from Google Cloud's Gemini Enterprise Agent Platform (formerly Vertex AI), which offers Gemini inside Google Cloud projects with enterprise controls.

Models. Google's documentation lists the Gemini 4 family (Carbon, Argon and Hydrogen), Gemini 3.x Flash and Flash-Lite models, image-generation models, text-to-speech models, Gemini Embedding and specialised models such as Gemini Decision 1 for the Decisions API. Preview and experimental models usually carry tighter rate limits.

Who it is for. Developers, startups and teams that want quick API-key access to Gemini models, a free tier for prototyping, and an easy path from AI Studio experiments to production code.

Pricing and limits. New projects start on a Free tier with lower limits, and some newer models are not available on it. Paid use requires linking a billing account; new users default to Prepay, which needs a minimum credit purchase, and requests stop with an HTTP 402 error when the prepaid balance reaches zero. Projects move from Tier 1 to Tier 2 and Tier 3 based on cumulative Google Cloud spend and time since the first payment. Each tier has its own rate limits (requests per minute, input tokens per minute and requests per day, applied per project rather than per API key), a monthly spend cap for the billing account, and in some cases a spend-based rate limit over a rolling 10-minute window. The Batch API costs 50% of standard pricing with a 24-hour target turnaround and separate limits, and Flex and Priority (preview) tiers trade price for latency. Current per-model limits are shown in AI Studio.

Key features. Long-context multimodal models, function calling, structured output, context caching, Grounding with Google Search, the Interactions API, Batch API, project-level spend caps and usage dashboards in AI Studio.

Alternatives. The OpenAI API, the Claude API from Anthropic, Gemini Enterprise Agent Platform for Google Cloud-managed access, and multi-provider routers such as OpenRouter.

Official site: https://ai.google.dev/gemini-api/docs

Official site