Skip to content

ModelsPerplexityUpdated

Perplexity releases Decider v1.3 open weights under Apache 2.0: decision models that classify, score and locate text

Decider v1.3 at a glance

Perplexity Decider v1.3 open weights arrived on November 12, 2026: the 27B decision model and its 4B Flash version return probabilities instead of text, add Locate span extraction, and are free to download under Apache 2.0.

Ask your AI about this story

Use this story in your business: Perplexity releases Decider v1.3 open weights under Apache 2.0: decision models that classify, score and locate text (Perplexity). Ask your AI to explain what happened, how it could affect your business and how to put it to work. It also weighs price, speed, intelligence and the lab behind it. The link to the story is included.

What changed with Decider v1.3

Perplexity released Decider v1.3 and Decider Flash v1.3 on November 12, 2026, the newest versions of its open-weight decision models. A decision model does not generate text: given an input and a set of questions, it returns a probability for each answer, so applications can classify, route, rank and evaluate content in a single forward pass. Decider v1.3 is built on Qwen3.8-27B, and Decider Flash v1.3 is distilled from it onto Qwen3.5-4B. Both models are on Hugging Face under the Apache 2.0 licence, on OpenRouter and in the Perplexity Decisions API. The new Locate feature extracts verbatim text spans, with character offsets and confidence scores, for named-entity recognition, personal-information detection and extractive question answering.

In Perplexity's own evaluation, Decider v1.3 reached 91.6% mean accuracy across 63 classification tasks, 6.54 percentage points above the production classifiers it replaces, and led TypeSafe's Jev 1.14 across seven domain suites, by up to 12.9 points in finance. Perplexity also reports using Decider internally to audit agent security logs: in one day it read 1 billion tokens across 7,000 agent sessions and identified flaws that led to 14 security-rule fixes, 6 of them approved and shipped within two days. As of November 13, 2026, Artificial Analysis had not published a page for Decider v1.3, so this item carries no Artificial Analysis metrics.

Decider v1.3 in numbers

Values are copied from the linked source on the date shown. They are not IA Compa measurements.

Mean accuracy, 63 classification tasks (Perplexity's evaluation)as of · Source for Mean accuracy, 63 classification tasks (Perplexity's evaluation): perplexity.ai (opens in a new tab)
91.6%

Where to compare Decider v1.3

Independent leaderboards and official resources for comparing this model yourself. Rankings change often; each site shows its own current data.

Why Decider v1.3 matters for businesses

Why it matters: the classify, route and tag work inside business automations no longer needs a full generative model call. Because the Decider v1.3 weights are Apache 2.0, a business can self-host a decision model for lead triage, content moderation, document tagging or personal-data detection and pay only for its own compute, while the Flash version fits high-volume routing and filtering. For agencies, that turns intake sorting, support routing and compliance screening into cheaper, faster steps inside the agent workflows they already sell.

Put it to work: AI services

Get AI News by email

Verified AI stories, each with its official source, on Monday and Thursday mornings. Add breaking news to hear about the biggest ones as soon as we publish them.

Every story links its official source

Free, in English or Spanish

Unsubscribe with one click, any time

What to send

Decider v1.3 updates since publication

Dated changes since publication. The original article text above is kept as written on its date.

  1. Ranking (Updated)

    Ranking update (November 16, 2026): Liquid AI's d1.2-ultra took #1 on Artificial Analysis's Decision Index v2.1 with 74.8, leading 24 dedicated decision models. On AA's measurements, general-purpose LLMs still score higher on the same index (Claude Haiku 5.5 Max 77, GPT-6 Luna Max 76).

    artificialanalysis.ai (opens in a new tab)

Sources for this story

  1. Official
  2. Official

Last verified: . IA Compa summarizes public announcements for business readers. We are not affiliated with the companies mentioned. Always confirm details, pricing, and availability with the official source.

  • #perplexity
  • #decider
  • #decision-models
  • #open-weights
  • #apache-2-0
  • #classification
  • #locate
  • #qwen

Want this working in your business?

We turn new AI capabilities into agents, voice receptionists, and automations you can actually use.