Skip to main content

11 model families

Google models and APIs

Explore Google model families on spicyapi. Compare supported tasks, review current pricing and open the model playground.

All model creators

About Google

Google's Gemini families cover the cheap, high-frequency end of the text catalogue. All five publish a single chat task in the OpenAI chat shape, with tool calls and JSON output, so they slot into an existing client without a separate SDK.

Gemini 3.7 Flash and 3.5 Flash-Lite both read up to 1,048,576 tokens and return up to 65,536, with Flash-Lite starting at minimal reasoning effort for the cheapest possible turn. Gemini 3 Flash Preview takes a 256,000-token window and is aimed squarely at high-frequency short turns. Gemini 3.1 Pro Preview reads a million tokens and answers in up to 32,768, which suits a long read with a compact conclusion. Gemini 2.5 Flash is the settled older lane, with a million-token read and a 65,536-token reply.

Choose by the shape of the turn rather than by version number: a long read with a short answer, a short read repeated thousands of times, and a genuinely long reply are three different jobs. Filtering behaviour varies across this line, so read the policy tier on the family page before routing anything sensitive to it.

Models from Google

Choose a model family to review its supported tasks, input options and current availability.

From a model to your API request

Open a model, select a task and configure its inputs in the playground. The API tab provides the request format; the displayed price reflects your selected options.

API documentation