Inception: Mercury Decide (free)

Inception: Mercury Decide (free) is a structured decision model from Inception, published under the model id inception/mercury-decide:free. It sits in the decisions category rather than the general chat category, which tells you what it is for: it is built to answer specific, well formed questions rather than to hold an open ended conversation. Inception serves it as a System One endpoint, a name that signals fast, direct responses of the kind you would use inside a larger pipeline rather than at the front of a chat product.

The model takes text as its only input and returns decisions as its output. According to Inception, you send a state along with typed questions, and it returns a choice, a score, or a yes/no answer. That output shape is the defining feature: instead of free form prose, you get a structured result you can branch on in code. The context window is 65536 tokens, and the maximum output is 58982 tokens. It supports no request parameters beyond the basics of the call itself, so there is no temperature, top_p or similar knob to tune.

The practical uses follow directly from that design. Anywhere your application needs to make a repeatable judgement over supplied context, Mercury Decide is a candidate: routing or triage decisions, scoring an item against criteria you describe in the state, or answering yes/no questions about a body of text. Because the input is text and the output is a decision, it fits naturally as a step inside an agent loop, a moderation or classification stage, or a workflow where a later component consumes the choice or score. The large context window means you can include substantial supporting material in the state without splitting it across calls.

You call it through api14's OpenAI-compatible API using the model id inception/mercury-decide:free, so existing OpenAI SDKs and clients work with only a base URL and key change. Because the model lists no supported parameters, keep your request minimal: the model id, your messages containing the state and the typed questions, and nothing else. Do not assume sampling controls will be honoured, and treat the response as the decision payload rather than conversational text.

One thing to weigh is the free designation and the absence of tunable parameters. If your task needs creative writing, open ended dialogue or fine control over sampling behaviour, a general purpose chat model will fit better. Mercury Decide is at its strongest when the question is narrow, the answer is discrete, and you want that answer returned in a form your code can act on directly.