Perplexity: Decider V1.2

Perplexity: Decider V1.2 is a decision model from Perplexity, released by the provider on 2026-10-16. It is a 27B checkpoint that succeeds Decider V1.1 27B and keeps the same API contract, so existing integrations built against the earlier version should carry over without structural changes. It sits in the decisions category, which distinguishes it from the text generation models you may already be routing through the gateway.

The model accepts text and images as input. Its output modality is decisions rather than prose: instead of writing an answer, it reads content passed as state and returns a decision. The context window is 262144 tokens, and the maximum output is 235929 tokens. That combination of a large input window with a large decision output allowance means you can hand it substantial material, including images alongside text, and receive a structured decision in return.

Practical use follows from that shape. Because it consumes state and emits decisions, it fits workflows where you need a verdict, classification, routing choice or selection made over a large body of supplied content rather than a written explanation of it. Mixed text and image input makes it usable when the material under consideration includes screenshots, scanned documents or other visuals alongside written context. The long context window suits cases where the decision depends on a lot of background at once, and the generous output allowance means the decision itself can be large.

You call it through api14's OpenAI-compatible API using the model id "perplexity/pplx-decider-v1.2". Because the endpoint follows the OpenAI request format, you can use existing OpenAI client libraries and point them at api14. Note that this model lists no supported parameters, so there are no additional request options to set beyond the standard request structure; check the live price table on the page for current rates before you commit to a workload.

One thing to weigh before adopting it: the output is a decision, not text. If your application needs generated prose, explanations, summaries or conversational replies, a text output model is the better fit, and Decider V1.2 will not serve that need. It is the right choice when the deliverable is the decision itself, particularly when the input is large or multimodal.