Meta: Muse Spark 1.2 Contributor

Meta: Muse Spark 1.2 Contributor is a reasoning model from Meta, offered as a contributor tier aimed at developers who want to start building at lower cost. It sits within the Muse Spark family, and Meta positions this tier as meaningfully cheaper than Muse Spark while keeping the reasoning behaviour that defines the line. It is a text model in the sense that text is its only output, though it accepts more than text as input.

On the input side, the model accepts text, images, video and files, and it produces text output. Its context window is 1,048,576 tokens, and the maximum output is 943,718 tokens. Because it is a reasoning model, it can expose its reasoning process, and the supported parameters include include_reasoning, reasoning and reasoning_effort, which let you control whether reasoning is returned and how much effort is spent on it. Generation behaviour is tunable through max_tokens, temperature, top_p, top_k and repetition_penalty, and the model supports response_format and structured_outputs for constraining output shape. It also supports tools and tool_choice, so it can call functions you define.

Those capabilities point to a few practical patterns. The very large context window suits work where you need to keep a long document, a codebase, a transcript or a set of files in view at once, rather than chunking and stitching results. Mixed input modalities make it a fit for tasks that combine a written question with an image, a video clip or an attached file, such as summarising a recorded session, extracting details from a screenshot, or answering questions grounded in a document you supply. Tool calling and structured outputs make it usable as the reasoning step inside an agent or pipeline, where it decides which function to call and returns a predictable JSON shape. Reasoning controls let you dial effort up for harder problems and down for routine ones.

You call it through api14's OpenAI-compatible API using the model id "meta/muse-spark-1.2-contributor". Because the interface follows the OpenAI chat completions shape, existing OpenAI SDKs and clients work with a change of base URL and API key. The request parameters it supports are include_reasoning, max_tokens, reasoning, reasoning_effort, repetition_penalty, response_format, structured_outputs, temperature, tool_choice, tools, top_k and top_p. If you want reasoning traces back, set include_reasoning; if you want a fixed output schema, use response_format or structured_outputs; if you want the model to call your functions, pass tools and tool_choice.

One thing to keep in mind is that output is text only, so if you need generated images, audio or other modalities, this model is not the right fit. It is also a contributor tier, which Meta describes as a lower cost entry point into the Muse Spark family, so if you need the full Muse Spark model rather than the contributor variant, that is the alternative to consider.