pplx-decider-v1-27b

Perplexity opened its Decisions API on October 1, 2026, running on pplx-decider-v1-27b: an open-weight decision model of ~26B parameters fine-tuned from Qwen3.8-27B, released on Hugging Face under Apache 2.0. Like the rest of the new decision-model wave, it never generates text: it reads state (text, JSON, images) and returns typed probabilities — yes/no, multiple choice, ordered score — in a single non-autoregressive forward pass.

The context window is 262,144 tokens, and API pricing is $0.04 per million input tokens with output free — the cheapest managed decision API at launch. Perplexity's own 11-benchmark, 7,210-sample panel scores it 85.71% overall against TypeSafe Jev's 84.51%, with the widest gap on RAGTruth (88.80% vs 77.27%); Jev still leads on WinoGrande (90.70% vs 83.30%) and BBH (94.27% vs 82.80). Those figures come from the company — its docs publish no benchmark table — so treat them as vendor claims. A v1.1 checkpoint followed on October 7 at $0.02 per million input tokens and topped an independent third-party bench at 80/82 correct.

Provider-reported (11 benchmarks, 7,210 samples): 85.71% overall vs Jev 84.51%; leads RAGTruth 88.80 vs 77.27, trails on WinoGrande and BBH. Independent: none for v1. Third-party bench (Oct 10): Decider V1.1 80/82 correct, 0.445 s median.