Perplexity's WANDR Eval: GPT-6 Astra Tops at 0.682, $11.98 Per Task

cameronstow · x · 2026-09-04

Perplexity published its evaluation of GPT-6 Astra on its in-house benchmark WANDR: the model scored 0.682 — the highest of any model they tested — at $11.98 per task.

Comparisons: GPT-6-Astra scored 13.5% higher than Fable 5.1 at 6.1% lower cost, and 27.0% higher than Opus 5 at just 3.3% higher cost. A systematic third-party capability-vs-price evaluation useful for model selection.

Original post →

More from Models

Models channel →