As of: 2026-11-19 22:11 CET. Researched and written by AI agents (Claude Opus 5.5 in Claude Code), with a human editor. How we work: https://postcutoff.com/about/ Canonical page: https://postcutoff.com/m/pplx-decider-v1-27b/ # pplx-decider-v1-27b Perplexity, pplx-decider family. Status: current. Type: language model. Released: 2026-10-01. Training cutoff: not published. ## How to call it | Provider | Model id | Endpoint | Docs | |---|---|---|---| | Perplexity API | `pplx-decider-v1-27b` | https://api.perplexity.ai/v1/decisions | https://docs.perplexity.ai/docs/decisions/quickstart | | Hugging Face | | | https://huggingface.co/perplexity-ai/pplx-decider-v1-27b | ## Pricing - Input: $0.04 per 1M tokens - Output: $0 per 1M tokens - Unit as published: per 1M tokens (output tokens free; Decisions API) - Source: https://docs.perplexity.ai/docs/decisions/quickstart - Last checked: 2026-10-03 ## Specs - Input: text, image - Output: text - Context window: 262,144 tokens - Max output: not published - Open weights: yes - Licence: apache-2.0 ## What it has not seen - No published training cutoff. Counted from its release (2026-10-01): at least 1,298 AI events since (291 major, 57 historic), as of 2026-11-19. - Everything since September 2026: https://postcutoff.com/since/2026-09/ ## What stands out - Decision model (probabilities over fixed answers): Returns a probability distribution over the allowed options for 1–128 typed questions per request instead of generating text; Perplexity reports 85.71% on its 11-benchmark panel vs 84.51% for Jev. Source: https://huggingface.co/perplexity-ai/pplx-decider-v1-27b ## Notes Fine-tuned from Qwen3.8-27B. Context is the API's 262,144-token input limit. Benchmarks are Perplexity's own; Jev still leads on 4–6 of 11 tests. Open-weights decision model behind Perplexity's Decisions API. Send a request with your content plus typed questions to `POST https://api.perplexity.ai/v1/decisions`; output tokens are free. Model page: https://postcutoff.com/m/pplx-decider-v1-27b/