Google opens Gemini Pro API with a generous free tier

Just days after that controversial demo video made waves across the tech world, Google is officially putting Gemini into the hands of real builders.

The search giant just released Gemini Pro and Gemini Pro Vision for developers and enterprise customers, opening up API access alongside an aggressive set of developer tools and enterprise updates.

Generous free tiers and multi-language SDKs

For developers eager to start building, the fastest path is Google AI Studio, a free web-based prototyping environment.

Sign in with a standard Google account, grab an API key, and you get an eye-opening rate limit of 60 requests per minute at zero cost. Google claims that is roughly 20 times more generous than competing free tiers on the market right now.

The initial Gemini Pro model features a 32K context window, native support for function calling, embeddings, semantic retrieval, and grounding. It supports 38 languages across more than 180 countries and territories.

Google also rolled out SDKs for Python, Android Kotlin, Node.js, Swift, and JavaScript. Alongside the text-only Gemini Pro, developers can also access Gemini Pro Vision, a multimodal endpoint that accepts text and image prompts and generates text.

There is one small catch to the free tier. Your prompts and responses may be reviewed by trained human reviewers to improve Google products, though the data is stripped of account credentials. If you require strict privacy, enterprise pipelines are the way to go.

The enterprise stack gets a complete makeover

For enterprise workloads requiring total data sovereignty, Gemini Pro is available through Google Cloud's Vertex AI.

On Vertex AI, teams can customize, tune, and distill models using corporate data. Google explicitly guarantees that customer inputs and outputs are never used to train its foundation models.

Google also introduced several companion models and platforms across Vertex AI.

Imagen 2, DeepMind's newest text-to-image diffusion model, is now available with sharper detail and significantly improved text rendering for brand logos and signs. Google also launched MedLM, a family of healthcare-specific models built on Med-PaLM 2, with plans to integrate Gemini architecture soon.

Meanwhile, Duet AI for Developers and Duet AI in Security Operations have officially reached general availability.

Gemini Pro remains free within rate limits across AI Studio and Vertex AI until early next year, after which Google will transition to character and image-based pay-as-you-go pricing.

Why it matters

Google is playing a classic platform adoption playbook here.

For everyday builders, that 60 RPM free tier makes Gemini Pro an absolute no-brainer to test. You can prototype multimodal apps, test prompt grounding, and build automated workflows without entering a credit card. It dramatically lowers the barrier to entry for solo founders and smaller teams.

For the AI landscape, this is a direct shot across OpenAI's bow. Google is not waiting around for Gemini Ultra next year; it is mobilizing the developer layer today to erode GPT-3.5's dominance in high-volume, cost-sensitive use cases. Leveraging its custom TPU infrastructure lets Google sustain an aggressive pricing strategy that few competitors can easily match.

We think launching a cohesive ecosystem on day one—complete with multi-platform SDKs, rapid browser prototyping, and seamless migration into Vertex AI—was a very smart move. It transforms what could have been a routine API drop into a full-funnel developer acquisition pipeline.

The only real risk is model performance in the wild. If early production feedback reveals that Gemini Pro stumbles on complex multi-step reasoning compared to rival offerings, developers may take advantage of the free tier for basic experiments while keeping their mission-critical logic parked elsewhere.