pplx-decider-flash-v1.3

by Perplexity · 4B parameters

Current new · 5d budget open weights cheap tier

Email me when pplx-decider-flash-v1.3 gets cheaper

Availability: Sold by Perplexity through its Decisions API (POST /v1/decisions); any Perplexity API key works. Requests that name pplx-decider-flash-latest are served by this model and bill at the same $0.01. Input must stay under 256K (262,144) tokens per request. The weights are on Hugging Face under Apache 2.0.

Today's price · per 1M tokens

Input

$0.010

per 1M tokens

Output

per 1M tokens

Blended

$0.007

blended $/1M — 3:1 weighted input:output

How this price is scoped: Perplexity's own pricing page lists pplx-decider-flash-v1.3 at $0.01 per 1M input tokens, output tokens Free, per-request fee None, half the $0.02 of pplx-decider-v1.3 and the same rate as flash-v1.2. You pay for input only: the content you send, every question and any images all count as input, so a request with 600 input tokens costs $0.000006. The $0 output here is Perplexity's published rate, not a missing number. With web grounding on, each web search or URL fetch adds $0.001.

Source: official Perplexity pricing · read Nov 13, 2026 MODELPRICEWATCH.COM · 2026-11-14

Price receipt

We read Perplexity's own pricing page on Nov 13, 2026 and found pplx-decider-flash-v1.3 listed with a price on the same row — that read is where the number above comes from, recorded as a new model. We keep the page text we read; its content hash is 3a70ddc814.

Overview

Perplexity's small decision model, the 4B sibling of pplx-decider-v1.3 on the same Decisions API. It reads text, JSON or images plus your questions and returns a probability for each answer, not generated text. v1.3 is a new build of pplx-decider-flash-v1.2 on Qwen3.5-4B. It adds `locate` questions, which return matching text spans with offsets and scores. It also answers all of a request's questions in one pass instead of one at a time, so the input is billed once per group rather than once per question. It trades some accuracy against the 27B model for lower cost and latency. $0.01 per 1M input tokens; output is free and there is no per-request fee.

Run it yourself