Google: Gemini Pro Latest

Google: Gemini Pro Latest is a text model from Google. The model id on api14 is ~google/gemini-pro-latest, and its provider description states that it always redirects to the latest model in the Gemini Pro family. That means you are not pinning a specific version: requests to this id follow whatever Google currently serves as the newest Gemini Pro release, so behaviour can change over time without any change on your side.

The model accepts audio, file, image, text and video as input, and it produces text only. Its context window is 1,048,576 tokens, which is large enough to hold very long documents, long transcripts or a mix of text and media in a single request. The maximum output is 65,536 tokens, so responses can be long as well. Because input is multimodal, you can send images, audio, video or files alongside your instructions rather than converting everything to text first.

Those capabilities point to a fairly wide set of practical uses. Long context plus file and image input suits document analysis, summarising large reports, and question answering over a corpus you supply in the prompt. Video and audio input make it a candidate for meeting transcription and analysis, media review, or extracting structured information from recordings. Text output with structured outputs and response_format support fits pipelines where you need JSON or another fixed shape back, and tool calling lets the model drive function calls in an agent or workflow. Reasoning support is useful for tasks that benefit from a deliberate step before the final answer.

You call it through api14's OpenAI-compatible API using the model id ~google/gemini-pro-latest. The supported request parameters are include_reasoning, max_tokens, reasoning, reasoning_effort, response_format, seed, stop, structured_outputs, temperature, tool_choice, tools and top_p. In practice you set max_tokens to bound the output, use temperature and top_p to control sampling, use seed and stop for reproducibility and termination, and use tools with tool_choice when you want the model to call functions. For structured replies, response_format and structured_outputs are the relevant parameters, and reasoning, reasoning_effort and include_reasoning let you control and surface the model's reasoning.

One thing to keep in mind is that this id is a moving target. Since it redirects to the latest Gemini Pro model, the underlying model can be updated by Google at any time, which may affect output style, latency or behaviour on edge cases. If you need stable, reproducible behaviour across deploys, a pinned model version is usually the better fit. Also note the output modality is text only, so if you need generated images or audio back, this model is not the right choice.