Skip to main content

PerplexityDecision modelAvailable

pplx-decider-v1.3

PopularityRank 26 of 39 decision models

28out of 100first value

usage data only, range 14 to 42

Summary

pplx-decider-v1.3 is a Perplexity decision model based on Qwen3.8-27B. It processes text, JSON state and images, returning structured decisions with probabilities. Its Locate operation extracts text spans and their character offsets.

Perplexity offers the model through the Decisions API and as downloadable Apache 2.0 weights. The Decider client has a configurable max_length setting that defaults to 32,768 tokens. The text backbone is configured for 262,144 positions, and OpenRouter lists a 262,144-token context for its offering.

Popularity compared

How often the model is searched for, used and mentioned, compared with every model of its category.

Popularity index

0 to 100, higher is more popular

20 of 42 models

As of November 2026 · Source: gradually.ai popularity index

Over time

Weekly values from November 15, 2026 to November 20, 2026

pplx-decider-v1.3pplx-decider-flash-v1.3 30Kanana-2-v-decide-3b-2611 29Decider 0.8B 28Intern-Decision-0.8B 27

What is gaining popularity right now, across models, vendors, harnesses and AI subscriptions, is collected in All AI trends.

All specifications

Architecture

Hugging Face checkpoint
perplexity-ai/pplx-decider-v1.3
Source: Hugging Face checkpoint
Model class
Multimodal decision model; returns structured decisions rather than free-form text.
Source: Model class
Base model
Qwen3.8-27B
Source: Base model

Input and output

Input
Text, JSON, and images as state
Source: Input
Output
Structured decisions with probabilities and Locate extractions
Source: Output

Watermarking and provenance

AI watermarking refers to invisible signals embedded in the output. Visible marks and provenance metadata are listed separately. Unknown means there is insufficient evidence for that particular field.

AI watermarking
Unknown
no source
Watermarking technique
Not known
no source

Licence and access

Access
Perplexity confirms the Decisions API, Hugging Face, and OpenRouter as access routes. The announcement does not list a token price for the Perplexity API.
Source: Access
Shutdown
None announced
no source

Further details

Indexed backbone parameters
26,085,330,160 BF16 parameters in the indexed backbone shards. Readout and Locate are supplied as additional components.
Source: Indexed backbone parameters
Client default for max_length
The configurable max_length setting of the Decider client defaults to 32,768 tokens.
Source: Client default for max_length
Text backbone position limit
The model configuration sets text_config.max_position_embeddings to 262,144.
Source: Text backbone position limit
Training sequence length
Training sequences of up to 30,720 tokens.
Source: Training sequence length
OpenRouter context offering
OpenRouter lists a 262,144-token context for its offering.
Source: OpenRouter context offering
New extraction capability
Locate extracts values and their text spans with character offsets and confidence scores.
Source: New extraction capability
Weights and license
Apache License 2.0. The weights are publicly downloadable.
Source: Weights and license