Perplexity Releases WANDR Research Benchmark
denisyarats · x · 2026-07-15
Perplexity devs introduced a new research benchmark called WANDR:
- Scale: 500 tasks
- Goal: Evaluating "wide research"—not just finding a few search results, but locating all relevant results and providing evidence for each.
- Conclusion: Simple browse/search benchmarks are becoming saturated; the real challenge lies in these broad-coverage research tasks.
They also noted that the Search as Code architecture within the Perplexity Agent API performs exceptionally well on WANDR. It allows the model to design a research plan first, then scale the research through deterministic execution without overwhelming the context window.
The post also shared links to:
- The wide-research preset
- The WANDR research article
- The GitHub repository
Related event: Perplexity open-sources internal research benchmark WANDR(9 posts)→
More from coding & agent
- Building a Secure AI Agent Gateway: Self-Hosting OAuth for Multiple SaaS Apps — Defiant_Cod_2654 · 2026-07-22
- Rowboat launches as an open-source, local-first AI coworker with memory — ycombinator · 2026-07-22
- Scoble says AI “loops” really means long-running multi-agent workspaces — Scobleizer · 2026-07-22
- Kimi Code opens a waitlist as Moonshot rolls out its coding product — Fabulous_Bonus_8981 · 2026-07-22
- Open-source runtime lets each repo define its own AI code reviewer — ibabufrik · 2026-07-22
- Indie Dev Asks: What's Actually Broken in Your AI Agent's Memory Today? — AcceptableTime7937 · 2026-07-22