Perplexity Adds Claude Fable 5.1 to Computer for Pro and Max.

Perplexity is adding Claude Fable 5.1 to Perplexity Computer, giving Pro and Max subscribers another model option for completing research and other multi-step tasks. The company says the model delivered both better results and lower operating costs than its predecessor in a recent evaluation.

The update was announced on LinkedIn. According to Perplexity, Claude Fable 5.1 ranked first in its August WANDR evaluation, recording a score of 0.601 at a cost of $12.76 per task.

Those figures represent a 21% higher evaluation score and a 37% lower cost per task compared with Fable 5, Perplexity said. That combination matters for Computer because model selection is not only about raw capability. For workflows requiring repeated browsing, analysis and content generation, the cost of each completed task can determine whether an automation is practical at scale.

Perplexity did not provide further information in the announcement about the WANDR evaluation, including the number or types of tasks tested, scoring methodology, failure rates or how the $12.76 cost was calculated. The result should therefore be treated as a Perplexity-reported benchmark rather than an independent comparison.

Still, the direction of the reported change is notable. A 21% score improvement suggests the model completed Perplexity’s test tasks more successfully, while the 37% reduction in cost indicates it required fewer resources or used them more efficiently. The announcement does not say whether subscribers will see those savings directly, and the per-task figure should not be confused with subscription pricing.

For small businesses and builders, the practical questions will be whether Claude Fable 5.1 improves reliability on real work such as competitive research, document synthesis and structured content production. A lower benchmark cost could also help Perplexity run more complex workflows within its paid plans, although the company has not announced changes to usage limits.

The addition continues Perplexity’s strategy of expanding the model selection available through Computer. It follows other model releases, including GLM 5.3 for long-context research and DeepSeek V4 Pro for U.S.-hosted workflows. Perplexity has also been broadening the information available to those workflows through licensed business-data connectors.

Claude Fable 5.1 is available to both Pro and Max subscribers, according to the announcement. Perplexity did not specify whether users need to select the model manually, whether it can be chosen for every Computer workflow or if its availability varies by task.

Users evaluating it against Fable 5 should focus on completed outputs rather than benchmark scores alone. Useful comparisons include how often a task requires correction, whether sources are handled consistently, how well instructions survive across multiple steps and whether the model produces usable structured results. Those factors can outweigh small differences in speed or token costs for business-critical workflows.

The headline numbers give Claude Fable 5.1 a strong initial case inside Computer, but the brief announcement leaves several operational questions unanswered. Its value will ultimately depend on whether Perplexity’s reported gains carry over from WANDR into the varied, less predictable tasks subscribers run every day.

What is Perplexity Claude Fable 5.1?

Claude Fable 5.1 is a newly available model option within Perplexity Computer. Perplexity says it scored 0.601 in its August WANDR evaluation while costing $12.76 per task.

Who can use Claude Fable 5.1 in Perplexity Computer?

The model is available to Perplexity Pro and Max subscribers. The announcement did not provide details about other plans or any separate access requirements.

How does Claude Fable 5.1 compare with Fable 5?

Perplexity reports that Claude Fable 5.1 achieved a 21% higher WANDR score and a 37% lower cost per task than Fable 5. The company did not publish additional benchmark methodology or independent validation with the announcement.