# Perplexity AI releases decision model pplx-decider-v1-27b fine‑tuned from Qwen3.8-27B

Digest AI · Generative AI & Models · published 2026-10-02T00:35:39Z

Canonical: https://digestai.news/story/perplexity-ai-releases-decision-model-pplx-decider-v1-27b-finetuned-fr

## Summary

Perplexity AI has open‑sourced a new decision‑making model called **pplx-decider-v1-27b**. The model is a fine‑tune of the Qwen3.8‑27B base and was evaluated on eleven public benchmarks, with the best scores highlighted in the release notes. Accuracy figures are reported through the Perplexity API, though the article does not quote exact percentages.

The model runs on Python 3.12+ with a CUDA‑enabled GPU and requires roughly **49 GiB** of VRAM for the weight files plus additional working memory. Users can download the weights from Hugging Face and run inference via a simple script or the provided `uv` commands. Example code shows how to classify a support ticket and how to request a yes/no probability for urgency, with optional image input.

Perplexity AI positions the decider as a lightweight, open‑weight alternative for developers needing calibrated choice predictions without building a full‑scale LLM pipeline. The release includes a ready‑to‑run inference example and clear installation steps, making it immediately usable for downstream applications such as ticket routing or urgency detection.

## Key points

- pplx-decider-v1-27b is a decision model fine‑tuned from Qwen3.8‑27B and evaluated on eleven benchmarks
- The model needs about 49 GiB of GPU memory and runs on Python 3.12+ with CUDA
- Weights are hosted on Hugging Face and can be run via a short uv‑based inference script

## Why it matters

Open‑weight decision models let developers add calibrated classification to apps without the cost of large generative LLMs, expanding affordable AI tooling for many businesses.

## Sources

1. [Perplexity Decider 27B: Open weights decision model fine tune of Qwen3.8 27B](https://huggingface.co/perplexity-ai/pplx-decider-v1-27b) (huggingface.co, 2026-10-02, primary source)

Part of the developing story: [AI Embedding Race Heats Up](https://digestai.news/thread/linkup-research-releases-sparseup-149m-parameter-sparse-embedding-model) (3 stories)

## Cite

Digest AI, "Perplexity AI releases decision model pplx-decider-v1-27b fine‑tuned from Qwen3.8-27B", 2 October 2026, https://digestai.news/story/perplexity-ai-releases-decision-model-pplx-decider-v1-27b-finetuned-fr

---

Written by Digest AI's editorial model from the linked sources; the sources are the record. Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse
JSON: https://digestai.news/story/perplexity-ai-releases-decision-model-pplx-decider-v1-27b-finetuned-fr.json