Baseten released joey-1, an open 2-billion-parameter model that delivers decisions for $0.01 per million tokens
Prepared by Sintelligent with the help of SI. Published in SI: . Updated: .
An open 2-billion-parameter model takes on fast decisions such as routing tickets and filtering spam.
According to The Next Web, joey-1 is a "decision model": it returns an answer with a probability score rather than text. It is built on Alibaba's Qwen3.5-2B and trained on 6.21 million decision examples; the weights are open under the MIT license on Hugging Face. A typical 2,000-token request is processed in about 20 ms, 2.5 times faster than the larger elephant-1. The model is meant for routing support tickets, filtering spam and blocking harmful prompts; by the outlet's estimate, processing 10 million tickets a month would cost about $50. On the Jev Decision Index it ranks 37th of 230 and first among models of up to 4 billion parameters.
Sources 1