Perplexity lists its v1.2 and smaller Flash decision models on Hugging Face
Perplexity says Flash handles text and images with the same API as its 27B model, at about one-seventh the size.
TLDR
Perplexity says its v1.2 decision model can answer multiple questions about the same input in one request. Flash uses the same API at about one-seventh the size, with text and image support. Perplexity says Flash retains most of the 27B model’s Decision Index and vision scores in its evaluations. Both Hugging Face model cards say the repositories are private and require a token with access.
Combined views
862
1 Source, first seen ago