WANDR (Wide ANd Deep Research) (WANDR)
We show this table for reference; we do not rank on it.
Perplexity's wide-search research benchmark: 500 tasks in which a model must find tens to hundreds of entities, collect facts about each, and return them as cited structured records.
Benchmark score on WANDR — November 11, 2026
We compile the WANDR rows from provider self-reports. Claude Fable 5.2 leads the table at 82.1%, followed by Claude Opus 5.5 (72.3%) and Claude Fable 5.1 (68.7%). We do not use these results to rank models overall.
Claude Fable 5.2
Anthropic
Claude Opus 5.5
Anthropic
Claude Fable 5.1
Anthropic
3 modelsAgenticCurrentDisplay onlyUpdated November 11, 2026
Benchmark score table (3 models)
| Rank | Model / configuration | Score | Parameters (B) | Open / closed |
|---|---|---|---|---|
| 1 | Claude Fable 5.2Anthropic | 82.1% | Not reported | Closed |
| 2 | Claude Opus 5.5Anthropic | 72.3% | Not reported | Closed |
| 3 | Claude Fable 5.1Anthropic | 68.7% | Not reported | Closed |
Among the reported WANDR rows, Claude Fable 5.2 is first at 82.1%. The third row is 13.4 points behind. The broader top-10 range is 13.4 points, so the table still separates the published systems.
3 models have been evaluated on WANDR. The benchmark falls in the Agentic category. WANDR is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.
About WANDR
Year
2026
Tasks
500 wide-search research tasks
Format
Soft F1 (relaxed score)
Difficulty
Agentic search
Anthropic ran a modified version with a frozen offline search index and a Claude Opus 4.6 judge that re-fetches cited URLs; Perplexity's own version uses the live web and a GPT-5.4 judge, so these scores are not comparable with Perplexity's published numbers. The metric is soft F1 (the launch table calls it the relaxed score). BenchLM first added WANDR from the Claude Fable 5.2 launch and keeps it display-only.
Freshness and provenance
Version
WANDR 2026
Refresh cadence
Quarterly
Staleness state
Current
Question availability
Public benchmark set
BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.
Questions
What does WANDR measure?
Perplexity's wide-search research benchmark: 500 tasks in which a model must find tens to hundreds of entities, collect facts about each, and return them as cited structured records.
Which model scores highest on WANDR?
Claude Fable 5.2 by Anthropic currently leads with a score of 82.1% on WANDR.
How many models are evaluated on WANDR?
3 AI models have published results on WANDR in the BenchLM catalog.
Compare top models on WANDR
Know when it’s worth switching models
The model to choose, the cheaper alternative, and the release we would wait on.
Read a sample issueJoin 7,000+ subscribers.
One email each week. Unsubscribe anytime.