Perplexity Prices New Decision Models From $0.01 per Million Input Tokens

Perplexity released two Apache 2.0-licensed decision models with API input prices starting at $0.01 per million tokens, aimed at tasks that need classifications or choices instead of long-form answers.

The 27B pplx-decider-v1.2 costs $0.02 per million input tokens, while the 4B pplx-decider-flash-v1.2 costs $0.01. Output tokens are free. Both models are available through the Perplexity API and Vercel AI Gateway. Their repositories are private and require authorized access.

The models process text and images to return classifications, choices or scores. Documented uses include routing agent tasks, scoring, labeling and filtering. A shared-input format lets developers submit several questions about the same text or image in one request for a single inference pass.

Combining questions can reduce accuracy with the 4B model, so its default format uses one question per request. The models are based on Qwen3.8-27B and Qwen3.5-4B, respectively.

On Decision Index 0.3, the 27B model scored 67.75 when answering one question per request, compared with 62.30 for its predecessor in the same format. Its score was 68.09 with joint requests. The v1.2 training data included official training splits for some evaluated tasks; those splits were decontaminated against test sets.