Claude Opus 5 is now available in Perplexity and Perplexity Computer.
We evaluated it against six other models on WANDR. It outperformed all but Fable 5, while being 57% cheaper.
Perplexity's July 24 evalA repeatable test for AI quality — a set of tasks plus scoring — used the way software teams use test suites, because model output is too variable to judge by eyeballing.Full definition → positioned Claude Opus 5 as a cheaper alternative to the highest-scoring configuration in Computer. On WANDR, a test of evidence-backed research, Opus 5 scored 0.352 at $12.35 per task, versus Fable 5's 0.427 at $28.79. The advertised 57% cost reduction therefore accompanies a lower score. Opus 5 ranked second among the seven configurations shown, offering a specific cost-quality tradeoff within this evaluation. The announcement and chart provide the figures.
Terms in this piece · Glossary
eval — A repeatable test for AI quality — a set of tasks plus scoring — used the way software teams use test suites, because model output is too variable to judge by eyeballing.
