Perplexity's Decider V1.2 models now available on AI Gateway
Source
vercel.com
Author
Jerilyn Zheng
Date
Why it matters
Decider models return probabilities for typed questions, so you can route tickets, filter content or escalate uncertain cases to humans through the AI SDK with one call. The 4B Flash variant suits high-volume labeling.
Key takeaways · AI-distilled
Boolean questions return the likelihood of yes, Choice questions pick from categories the developer defines, Score questions grade against an ordered rubric, and several questions can be asked about the same input in one request.
The 27B model (perplexity/pplx-decider-v1.2) is positioned for nuanced judgments and detailed rubrics, while Flash (perplexity/pplx-decider-flash-v1.2) targets tasks with clear criteria.
The models can be called through the AI SDK's experimental_decide function, a Decisions API, or a TypeSafe-compatible API, and images are passed as base64 image_url objects embedded in JSON state.
According to Vercel, decision requests appear in AI Gateway request logs and count toward spend budgets, and inferenceRunning a trained model to get answers — the phase where AI is actually used, as opposed to trained.Full definition → is billed at provider pricing with no markup or platform fee, including under BYOK.
Terms in this piece · Glossary
multimodal — A model that works with more than text — reading images, audio, or video, and sometimes generating them too.
inference — Running a trained model to get answers — the phase where AI is actually used, as opposed to trained.
