decisionopenHand-written lesson

Perplexity

Decider V1.2

perplexity/pplx-decider-v1.2

A 27B decision model from Perplexity. Instead of generating text, it reads the content you pass as state and answers typed questions with probabilities your code thresholds.

RoleThe ticket router

$0.020

In / M

Free

Out / M

262K

Context

AA intelligence index
Not on listing
Inputs
text, image
Outputs
decisions
Tools
Not listed
Structured output
Not listed
Reasoning
Not listed
Weights
Open · perplexity-ai/pplx-decider-v1.2
Listed
Oct 16, 2026
Live from the OpenRouter listing ↗

01 · What it's good at

Use it for this.

Three jobs this model is built for, and the ones to give to something else.

01

Routing tickets, reviews, and screenshots to a team

02

Moderation where only confident hits are auto-hidden

03

Grading answers against your rubric

Skip it for

  • Writing, chat, or code
  • Copying a threshold from another decision model without re-testing

02 · Example tasks

What you'd actually send it.

Real inputs, not a benchmark. Paste one into the snippet below to try it.

You send

State: the ticket text. Questions: noul 'needs a human', choice of team, score urgency 1 to 3.

Why this model

Three decisions in one call, each a probability.

03 · Your stack

Where it sits: Gate & route.

In front of expensive work: decide which queue, which model, or whether a human needs to look, before anything writes text.

04 · Wire it up

One request. Paste and run.

Uses OpenRouter, so one key works for every model on this site. Set OPENROUTER_API_KEY and swap the prompt for one of the tasks above.

const res = await fetch("https://openrouter.ai/api/alpha/decisions", {
  method: "POST",
  headers: {
    Authorization: `Bearer ${process.env.OPENROUTER_API_KEY}`,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    "model": "perplexity/pplx-decider-v1.2",
    "state": {
      "ticket": "My checkout page shows a blank screen after I click Pay. I tried two browsers."
    },
    "questions": {
      "is_bug": {
        "type": "noul",
        "instructions": "Is the customer reporting a software defect?"
      },
      "team": {
        "type": "choice",
        "instructions": "Which team should own this ticket?",
        "criteria": {
          "payments": "Checkout, billing, or payment issues.",
          "account": "Login or profile issues.",
          "frontend": "Rendering or layout issues."
        }
      }
    }
  }),
});
if (!res.ok) throw new Error(`Decisions ${res.status}: ${await res.text()}`);
const { answers } = await res.json();
// A noul answer is the probability of "yes". Your code picks the thresholds.
const p = answers.is_bug.noul;
if (p >= 0.9) console.log("yes");
else if (p <= 0.1) console.log("no");
else console.log("send to a human", p);

Decisions is an alpha endpoint on OpenRouter. The TypeScript SDK exposes it as openRouter.alpha.decisions.create().

05 · Watch out for

Gotchas.

  • Probabilities differ between decision models. Measure thresholds on your own labeled data.

  • The Decisions API is an alpha endpoint on OpenRouter. Pin your SDK version.

From our review · 2026-10-18

You send a state (the ticket, the review, the image) and named questions. A noul question is yes/no and returns the probability of yes. A choice question returns a probability per option and a top pick. A score question rates an ordered rubric and returns a probability per level plus an expected score. Your code applies a threshold, for example auto-close if p(spam) > 0.95, otherwise send it to a person.

Perplexity also ships Decider Flash v1.2, a smaller sibling with the same question types, and a separate Orchestrator API that does not pick a label for you. The orchestrator routes a request to a healthy deployment of a model you already chose.

Perplexity said Decider v1 scores 85.71% across benchmarks, measured through its own API. Treat that as a vendor claim. For an independent cut, use Artificial Analysis's Decision Index. v1-27b weights were posted on Hugging Face, fine-tuned from Qwen3.8-27B; v1.2 itself was not verified as open on Oct 18, 2026.

Sources

Lessons are written from each model's listing and the linked docs. We don't publish our own benchmark numbers; the index above is Artificial Analysis, as shown on OpenRouter.

Keep learning

Same board, other models.