FinSearchComp Task 3: Complex Historical Investigation (FinSearchComp (T3))
We show this table for reference; we do not rank on it.
The hardest task of FinSearchComp, an open financial search and reasoning benchmark for end-to-end agents: multi-step investigations over historical market and company data that mirror a financial analyst's research workflow.
Benchmark score on FinSearchComp (T3) — November 13, 2026
We compile the FinSearchComp (T3) rows from provider self-reports. LongCat-2.5 leads the table at 63.4%. We do not use these results to rank models overall.
1 modelAgenticCurrentDisplay onlyUpdated November 13, 2026
Benchmark score table (1 model)
The one result on this page was reported by the lab that built the model.
| Rank | Model / configuration | Score | Who ran it | Parameters (B) | Open / closed |
|---|---|---|---|---|---|
| 1 | LongCat-2.5MeituanReported by Meituan | 63.4% | Reported by Meituan | Not reported | Open |
- Lab-reported
- Benchmark-run
- Independent
How to read this leaderboard
Operator receipt: 1 sourced row is currently displayable on this page; the leading published row is LongCat-2.5 at 63.4%.
Honest limit: Results depend heavily on the search tools and financial plugins available to the agent, which launch tables rarely describe.
About FinSearchComp (T3)
Year
2025
Tasks
Complex historical financial investigation questions (T3 subset of 635 total)
Format
Accuracy (%)
Difficulty
Multi-step financial search and reasoning
ByteDance Seed released FinSearchComp in September 2025 with 635 expert-annotated questions over global and Greater China markets, split into Time-Sensitive Data Fetching (T1), Simple Historical Lookup (T2) and Complex Historical Investigation (T3). Seventy professional financial experts annotated the tasks. Meituan reports LongCat-2.5 at 63.4 on T3 in its November 2026 model card, next to in-house runs of GPT-5.5 and GPT-5.6 Sol (62.2 each), Claude Opus 5 (64.0), GLM-5.3 (62.2), Kimi K3 (64.0) and DeepSeek V4.1 Flash (64.3). BenchLM stores the LongCat-2.5 row; the other figures are Meituan runs listed for context.
Freshness and provenance
Version
FinSearchComp (T3) 2025
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 FinSearchComp (T3) measure?
The hardest task of FinSearchComp, an open financial search and reasoning benchmark for end-to-end agents: multi-step investigations over historical market and company data that mirror a financial analyst's research workflow.
Which model scores highest on FinSearchComp (T3)?
LongCat-2.5 by Meituan currently leads with a score of 63.4% on FinSearchComp (T3).
How many models are evaluated on FinSearchComp (T3)?
1 AI models have published results on FinSearchComp (T3) in the BenchLM catalog.
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.