Free Models Router

Free Models Router is a text model from OpenRouter. Rather than a single model with fixed weights, it is a router: when you send a request, it selects free models at random from the models available to it, filtering for models that can serve the request. The provider describes it as the simplest way to get free inference. It is listed under the text category and is identified on api14 by the model id openrouter/free.

The router accepts text and image input and returns text output, so it can handle both plain language prompts and prompts that include images. It offers a context window of 200,000 tokens, which is large enough for long documents, extended conversations and sizeable code files in a single request. The maximum output length is not specified in the model details, so you should treat output size as dependent on the underlying model that handles your request rather than a fixed published figure.

Because the router picks from free models at random, the practical use cases are the general text and vision tasks those models cover. You can use it for chat and question answering, summarising long inputs that fit within the context window, drafting and editing text, extracting structured data through response formats and structured outputs, and building agents or tool calling workflows. Image input extends this to tasks such as describing images, answering questions about screenshots or diagrams, and combining visual context with written instructions. It suits prototyping, experimentation and low stakes workloads where you want inference without committing to a specific model.

You call it through api14's OpenAI-compatible API using the model id openrouter/free. Requests follow the familiar chat completions shape, so existing OpenAI client libraries work with a change of base URL and model id. The router supports a broad set of request parameters: temperature, top_p, top_k, min_p, frequency_penalty, presence_penalty and repetition_penalty for sampling control; max_tokens and max_completion_tokens for output length; stop for stopping sequences; seed for reproducibility; logprobs and top_logprobs for token level detail; response_format and structured_outputs for constrained output; tools and tool_choice for function calling; and reasoning, reasoning_effort and include_reasoning for models that expose reasoning behaviour.

One thing to keep in mind is that the router selects models at random, so the exact model behind a request can vary between calls. That makes behaviour less predictable than pinning a single model, and parameters such as reasoning_effort or structured outputs may be honoured differently depending on which model is chosen. If you need consistent outputs, reproducible behaviour or a specific capability set, a dedicated model is likely a better fit. If you want free inference across a range of text and image tasks and can accept that variability, openrouter/free is a straightforward option.