WideResearch
We show this table for reference; we do not rank on it.
A broad research-agent benchmark for open-ended information gathering, synthesis, and answer construction across wide search spaces.
Benchmark score on WideResearch — October 29, 2026
We compile the WideResearch rows from provider self-reports and secondary reports. Mercury 2 leads the table at 92.3%, followed by Hy4 preview (83.9%) and Ling 3.1 Flash (83.3%). We do not use these results to rank models overall.
Mercury 2
Inception
Hy4 preview
Tencent
Ling 3.1 Flash
InclusionAI
17 modelsAgenticCurrentDisplay onlyUpdated October 29, 2026
Benchmark score table (17 models)
| Rank | Model / configuration | Score | Parameters (B) | Open / closed |
|---|---|---|---|---|
| 1 | Mercury 2Inception | 92.3% | Not reported | Closed |
| 2 | Hy4 previewTencent | 83.9% | Not reported | Open |
| 3 | Ling 3.1 FlashInclusionAI | 83.3% | Not reported | Open |
| 4 | Qwen3.8 MaxAlibaba | 81.9% | Not reported | Open |
| 5 | Atria Dawn PreviewShanghai Artificial Intelligence Laboratory | 81.9% | Not reported | Open |
| 6 | Kimi K2.6Moonshot AI | 80.8% | Not reported | Open |
| 7 | Ornith-1.5-397BOrnith AI | 80.8% | Not reported | Open |
| 8 | dots3-note PreviewDots Studio | 78.9% | Not reported | Open |
| 9 | Claude Opus 4.5Anthropic | 76.4% | Not reported | Closed |
| 10 | Qwen3.6 PlusAlibaba | 74.3% | Not reported | Closed |
| 11 | Qwen3.5 397BAlibaba | 74.0% | Not reported | Open |
| 12 | Ling 3.0 FlashInclusionAI | 73.6% | Not reported | Open |
| 13 | Kimi K2.5Moonshot AI | 72.7% | Not reported | Open |
| 14 | GLM-5Z.AI | 69.8% | Not reported | Open |
| 15 | Ornith-1.5-35B-A3BOrnith AI | 67.8% | Not reported | Open |
| 16 | Qwen3.6-35B-A3BAlibaba | 60.1% | Not reported | Open |
| 17 | Ornith-1.5-9BOrnith AI | 59.5% | Not reported | Open |
Among the reported WideResearch rows, Mercury 2 is first at 92.3%. The third row is 9.0 points behind. The broader top-10 range is 18.0 points, so the table still separates the published systems.
17 models have been evaluated on WideResearch. The benchmark falls in the Agentic category. WideResearch is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.
About WideResearch
Year
2026
Tasks
Open-ended research tasks
Format
Multi-source research evaluation
Difficulty
Broad research-agent workflows
WideResearch evaluates whether a model can sustain a research process over multiple sources and branches rather than answering from shallow retrieval. BenchLM tracks it as a display-only browsing and synthesis benchmark.
Freshness and provenance
Version
WideResearch 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 WideResearch measure?
A broad research-agent benchmark for open-ended information gathering, synthesis, and answer construction across wide search spaces.
Which model scores highest on WideResearch?
Mercury 2 by Inception currently leads with a score of 92.3% on WideResearch.
How many models are evaluated on WideResearch?
17 AI models have published results on WideResearch in the BenchLM catalog.
Compare top models on WideResearch
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 6,100+ subscribers.
One email each week. Unsubscribe anytime.