We use cookies. Cookies and browser storage keep you signed in and remember your preferences. With your permission we also measure how the site is used. You can any time. See our cookie policy.
baseten/joey-1Runs on your balance at the prices above. New accounts start with free usage.
Settings
Baseten: Joey 1 is a decision model from Baseten, released on 29 October 2026. It comes from Base Labs, Baseten's research team, and is built on Qwen3.5-2B, a small 2 billion parameter base. Rather than generating free form prose, Joey 1 is designed around a single job: taking a state and a set of typed questions and returning decisions. That focus makes it a different kind of component from a general chat model, and it is the reason it sits in the decisions category rather than alongside text generation models.
The model accepts text and images as input. The state you give it can be plain text or JSON, and the questions you ask are typed: yes or no questions, multiple choice questions, and ordinal questions. Its output is decisions rather than prose, so you should expect structured answers to the questions you pose instead of an essay or a chat reply. The context window is 16,384 tokens, and the maximum output is 14,745 tokens, which is generous relative to the input size and leaves room for a large batch of questions and their answers in a single call.
The practical uses follow directly from that shape. Joey 1 fits workflows where something else has already gathered the state and you need a consistent judgement applied to it: classifying or routing an item, answering a checklist of yes or no questions about a record, picking one option from a defined set, or placing a judgement on an ordinal scale. Because it accepts images alongside text, the state can include visual material, so screenshots, scans or photos can be part of what the model reasons over. It is a natural fit as a scoring, gating or triage step inside a larger pipeline, where the surrounding system handles retrieval, tool calls and the final response, and Joey 1 supplies the decision.
You call it through api14's OpenAI-compatible API using the model id "baseten/joey-1". Because the interface follows the OpenAI chat completions shape, existing OpenAI SDKs and clients work with a change of base URL and model id. The model does not list any supported request parameters, so treat it as taking the standard request body without extra tuning knobs such as temperature or top_p. Send your state and your typed questions in the request and read the decisions from the response.
The main thing to weigh is the 16,384 token context window. If your state is a long document, a large JSON payload or a lengthy conversation history, it may not fit, and a model with a longer context would suit that case better. Joey 1 is also not the right choice when you need generated prose, since its output is decisions. Where the task is a bounded set of typed questions over a state that fits the window, it is built for exactly that.
import OpenAI from "openai";
const client = new OpenAI({
baseURL: "https://api14.com/api/v1",
apiKey: process.env.API14_API_KEY,
});
const result = await client.post("/decisions", { body: {
"model": "baseten/joey-1",
"state": "Customer: I was charged twice for my order and want my money back.",
"questions": {
"team": {
"type": "choice",
"instructions": "Which team should handle this message?",
"criteria": {
"billing": "Payments, refunds, invoices",
"support": "Product questions and bugs",
"sales": "Pricing and new purchases"
}
},
"urgent": {
"type": "noul",
"instructions": "Does this need a reply today?",
"criteria": {
"true": "Money, outage or legal risk",
"false": "Can wait"
}
}
}
} });
console.log(result.answers);
This community has no topics yet. Open the first one about Baseten: Joey 1.