Google: Gemini 3.5 Flash Lite

Google: Gemini 3.5 Flash Lite is a text model from Google, released on 21 July 2026. Google positions it as a high-efficiency model with upgraded agentic capabilities, and describes it as suited to subagents that execute focused tasks within complex, multi-agent workflows. It sits in the text category on api14 and is aimed at developers who need a capable model for well-scoped jobs rather than a general-purpose flagship.

The model accepts text, image, video, file and audio as input, and produces text as output. Its context window is 1,048,576 tokens, and the maximum output is 65,536 tokens. That combination means you can send very large inputs, including long documents or media alongside text, and still receive a substantial text response in a single call. Supported request parameters include temperature, top_p, max_tokens, stop, seed, response_format, structured_outputs, tools, tool_choice, reasoning, reasoning_effort and include_reasoning.

Those capabilities point to a few practical patterns. With tools and tool_choice available, the model can drive function calling and act as a subagent that picks up a narrow task inside a larger orchestration, which matches Google's own description. Structured outputs and response_format make it a reasonable fit for extraction, classification and other jobs where you need the reply in a fixed shape. The large context window suits long-document question answering, summarisation and analysis, and the mixed input modalities let you include images, audio, video or files in the same request when a task depends on more than text. The reasoning parameters give you a way to trade off deliberation against latency for tasks of varying difficulty.

You call it through api14's OpenAI-compatible API using the model id "google/gemini-3.5-flash-lite". Because the interface follows the OpenAI schema, you can point an existing OpenAI client at api14 and set that id, then pass the parameters listed above in the request body. The same id works for chat completions style requests, and the supported parameters cover the usual sampling controls plus tools, structured outputs and reasoning options.

Keep the output ceiling in mind: responses are capped at 65,536 tokens, so tasks that need a much longer single reply may call for a different model. The model is also text-only on the output side, so anything requiring generated images, audio or video falls outside what it returns. For work that is broad rather than focused, or that needs output modalities beyond text, another model on api14 may be the better choice.