Despite the benchmark drop, Perplexity's open weights system continues to lead medical evaluations, scoring 643 of 669 correct responses on the clinical System One board. Eight competitors, including models from Microsoft and Inception, failed to surpass that score as of Oct. 11. The model remains the second most used in its class on OpenRouter, where it is priced at $0.02 per million input tokens with free outputs.
Tuesday, October 20, 2026
Story
vLLM Decision 3.0 and Mapika Decider Tie for #1 on Decision Index
AI9 days ago
A new set of performance metrics for multimodal AI has triggered a change in leadership among the world's most accurate discriminator models. The Jev Decision Index 0.3.1 now places vLLM's Decision 3.0 (d3) and Mapika Decider 31B in a tie for the top spot, displacing Perplexity's pplx-decider-v1.1-27b to second place. The vLLM model's architecture was developed by Perplexity, and the update also saw AnyJev Gemma-4-31B become the first untuned model to exceed the performance of Jev, tying for fourth.
Looking for more context or analysis?
Tilt chat provides reliable, tailored insights, alerts, and automations.