Baseten releases open-weight joey-1 decision model
Baseten released joey-1, a 2B-parameter open-weight decision model that returns a label and calibrated probability instead of generating explanatory text. The Next Web reports that the model is built on Qwen3.5-2B, while Baseten Blog says it is available through Baseten Model APIs and with open weights on Hugging Face under the MIT license. Baseten reports 20 ms P50 end-to-end latency for a typical 2,000-token request and pricing of $0.01 per million tokens.
Joey-1 gives developers an open-weight option for high-volume classification and routing workloads where Baseten says joey-1 can combine direct decision outputs with low latency and per-token pricing.
Published
October 29, 2026
TL;DR
Eight days ago we released mammoth-1 (27B) and elephant-1 (9B), the #1-ranked decision models. Today we’re adding joey-1 (2B), our fastest and most affordable decision model, with a P50 latency of 20 ms, 5x faster than comparable decision model models.
Many LLM use cases don’t need generated text; they need decisions: