Google Lists Gemini 3.1 Pro Preview API Prices at $2 Input and $12 Output per Million Tokens

Google AI for Developers has listed Gemini 3.1 Pro Preview API pricing, naming model IDs and usage limits for the preview model in its official Gemini API documentation.

Official pricing for the preview model

Google AI for Developers’ Gemini Developer API pricing page lists Gemini 3.1 Pro Preview with the model IDs gemini-3.1-pro-preview and gemini-3.1-pro-preview-customtools. For paid API use with prompts up to 200,000 tokens, the page gives prices of $2 per 1 million input tokens and $12 per 1 million output tokens.

The same Google pricing document says different pricing applies for longer-context usage, Batch, Flex, and Priority modes. The excerpted source does not provide all of those higher rates, so the clearest comparable figure is the standard paid API price for prompts up to 200,000 tokens.

Context window and output limit

Google’s model documentation for Gemini 3.1 Pro Preview identifies gemini-3.1-pro-preview as the model name and lists a 1,048,576-token input limit. The same official model page lists a 65,536-token output limit.

Those figures align with Price Per Token’s catalog entry, which describes the model as having a roughly 1.0 million-token context window. Price Per Token says its Gemini 3.1 Pro Preview page was updated June 6 and lists $2 input and $12 output pricing per million tokens.

Features listed in Google’s model documentation

Google AI for Developers’ model page says Gemini 3.1 Pro Preview supports multimodal inputs. It also lists Batch API support, caching, Flex inference, function calling, and a custom tools endpoint.

The official pricing page separately identifies gemini-3.1-pro-preview-customtools, indicating that Google is documenting a custom-tools variant alongside the standard preview model ID.

Third-party catalog context

Price Per Token’s catalog page adds comparison-oriented fields, including provider routes and benchmark fields such as GPQA. Because those details come from a third-party catalog rather than Google’s official documentation, they should be treated as catalog metadata rather than Google-published product claims.

The most directly sourced facts are from Google AI for Developers: the preview model IDs, the $2 input and $12 output paid API prices per million tokens for prompts up to 200,000 tokens, the 1,048,576-token input limit, and the 65,536-token output limit.

Why it matters

API pricing and context limits determine the practical cost of deploying large-context models in applications such as retrieval, code analysis, document review, and tool-using assistants. Google’s documentation gives developers a baseline for estimating Gemini 3.1 Pro Preview costs, while also indicating that long-context and specialized service modes may use different rates.

Hero image prompt: Abstract editorial artwork of a large glowing token meter connected to cloud servers and document icons, with subtle numerical patterns and a clean technology-news style; no logos, no real product interfaces, no readable text.

Hero image alt text: Abstract illustration of API token pricing and long-context AI model usage represented by a glowing meter, servers, and document shapes.