Perplexity Open-Sources WANDR Benchmark for AI Research Capabilities.

@AravSrinivas· July 14, 2026 View original

Summary

Perplexity has open-sourced WANDR, an internal benchmark used to measure the research capabilities of AI systems, which they claim contributes to their cost-effective and high-performing research harness.

Perplexity has announced the open-sourcing of WANDR, a benchmark tool designed to evaluate the research capabilities of AI systems. The company attributes its strong performance and cost efficiency in AI research to robust internal evaluation processes, of which WANDR is a key component. By making this benchmark publicly available, Perplexity aims to contribute to the broader AI community's ability to assess and improve research-oriented AI models.

Why it matters

Professionals in AI development and research can use this new open-source benchmark to rigorously evaluate their own models, potentially improving performance and efficiency.

How to implement this in your domain

  1. 1Download and integrate the WANDR benchmark into your AI model evaluation pipeline.
  2. 2Compare your model's research capabilities against established benchmarks using WANDR.
  3. 3Analyze WANDR's methodology to refine your internal evaluation strategies.
  4. 4Contribute to the open-source WANDR project with feedback or improvements.

Who benefits

AI ResearchSoftware DevelopmentData ScienceAcademia

Key takeaways

  • Perplexity open-sourced WANDR, a benchmark for AI research capabilities.
  • WANDR helps evaluate AI model performance and cost-efficiency.
  • Open-sourcing promotes community-wide AI development and evaluation standards.
  • AI developers can use WANDR to improve their models.

Original post by @AravSrinivas

"Perplexity has the best (both on cost and performance) deep and wide research harness in Computer. One of the contributing factors is strong internal evals and benchmarks. Today, we're open-sourcing WANDR, the benchmark we use internally for measuring research capabilities."

View on X
Perplexity Open-Sources WANDR Benchmark for AI Research Capabilities.

Originally posted by @AravSrinivas on X · view source

Want to go deeper?

Turn these trends into skills with Learnijoy's hands-on AI & tech courses.

Explore courses