Google: Gemini 3.5 Flash

Google: Gemini 3.5 Flash is a text generation model from Google, listed on api14 under the model id google/gemini-3.5-flash. It sits in the text category and is described by its provider as a high-efficiency multimodal model that brings near-Pro level coding and reasoning at Flash-tier speed. If you are already working with Google's Gemini line, this is the variant aimed at workloads where latency and throughput matter as much as raw capability.

The model accepts text, image, video, file and audio as input, and returns text only. Its context window is 1,048,576 tokens, which is large enough to hold very long documents, extended conversation histories or sizeable codebases in a single request. The maximum output is 65,536 tokens, so responses can be long form when you need them to be. The knowledge cutoff is 2025-01-01, meaning anything after that date is outside its training data and should be supplied in the prompt if it matters.

Those capabilities point to a fairly wide set of uses. Because it takes images, video, audio and files alongside text, it fits document and media analysis tasks: summarising a recorded meeting, extracting structured data from scanned pages, or answering questions about a screenshot or clip. The long context suits repository-level code work, multi-document research and long-running agent loops where prior steps need to stay in view. Its provider notes optimisation for coding proficiency and parallel agentic execution, so tool-driven pipelines and code assistance are natural fits. Reasoning support also makes it reasonable for problems that benefit from a deliberate thinking step before the final answer.

You call it through api14's OpenAI-compatible API using the model id google/gemini-3.5-flash, which means existing OpenAI SDKs and clients work with only a base URL and key change. 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 that covers sampling control (temperature, top_p, seed, stop), output length (max_tokens), function calling (tools, tool_choice), structured JSON output (response_format, structured_outputs) and control over how much internal reasoning the model does (reasoning, reasoning_effort, include_reasoning).

One thing to keep in mind is that the model outputs text only. If your application needs generated images, audio or video as a result, this is not the model for that job, even though it can read those formats as input. The knowledge cutoff is also worth planning around: for questions about recent events, retrieval or prompt-supplied context will do more for accuracy than the model's own memory.