OpenRouter WideSearch (Search): leaderboard
OpenRouter's WideSearch run: the model must assemble a complete reference table from the open web, scored strictly — credit only when every required row and cell matches. Scored across search providers and turn budgets; each model takes its best configuration.
Metric: Best-configuration Exact Table Match (%). Source: openrouter.ai. Saturation forecast: Around 2032. 4 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.6 Luna (xHigh) | 22 | #79 (GPT-5.6 Luna) |
| 2 | GPT-5.6 Sol (High) | 21 | #22 (GPT-5.6 Sol) |
| 3 | Claude Opus 5 (High) | 20.2 | #16 (Claude Opus 5) |
| 4 | DeepSeek V4 Flash (0731) (High) | 20 | #84 (DeepSeek V4 Flash (0731)) |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=openrouter-widesearch-search · How It Works · Data refreshed daily, snapshot 2026-11-05.