Details
- Perplexity announced that Anthropic’s Claude Opus 5.5 is now available in Perplexity Computer at the Standard effort level.
- The release gives Computer users access to a new model option for web-based tasks and multi-step workflows.
- On Perplexity’s WANDR benchmark, Opus 5.5 achieved a score of 0.610.
- The model cost $4.13 per completed task in the cited evaluation.
- Opus 5.5 slightly outperformed Claude Fable 5.1 on the benchmark used by Perplexity.
- Perplexity says Opus 5.5 cost 67.6% less per task than Fable 5.1 in that comparison.
- Standard is Perplexity Computer’s everyday effort setting, intended to balance task quality, depth, and execution cost.
- The announcement expands Perplexity Computer’s model choices while positioning Opus 5.5 as a lower-cost alternative for comparable research and automation workloads.
Impact
The announcement increases competitive pressure in AI assistants by pairing a frontier Anthropic model with Perplexity’s task-execution interface rather than limiting it to standalone chat or developer access. If the WANDR figures generalize beyond Perplexity’s internal evaluation, the reported cost gap could make longer Computer workflows more economical and encourage users to trade up from cheaper models. The comparison should be treated as benchmark-specific because results depend on task mix, effort settings, and pricing assumptions.
