FinReasoning (Deep Insight) - Factuality and Contextualization: leaderboard
Metric: Rubric score (0-100): a judge score fused from DeepSeek-V3 and Qwen3-235B on three indicators scored 0 to 5 and weighted 40/30/30, comparing the model's open-ended analysis with an expert-verified reference answer (fact density, computational rigor, contextual analysis), on the 300 factuality items of FinReasoning's Deep Insight track (open-ended analysis of events in research-report excerpts); Chinese-language tasks built from A-share research reports, news and market data (2023 to 2025); zero-shot at temperature 0.1; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 19 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GLM-4.6 | 88.1 | #246 |
| 2 | Kimi K2 | 87.4 | #236 |
| 3 | Seed 1.8 | 87.2 | #136 |
| 4 | Claude Sonnet 4.5 | 87.1 | #138 |
| 5 | GPT-5 | 84.6 | #91 |
| 6 | DeepSeek R1 | 84.6 | #245 |
| 7 | Intern-S1 | 82.1 | #278 |
| 8 | DeepSeek V3 | 81 | #312 |
| 9 | Gemini 3 Pro | 79 | #77 |
| 10 | GPT-4o | 77.6 | #333 |
| 11 | Qwen 3 Max | 76 | #201 |
| 12 | Qwen 3 235B A22B | 63.8 | #304 |
| 13 | Qwen 3 32B | 61.7 | #424 |
| 14 | Qwen 3 8B | 51.1 | #667 |
| 15 | Llama 3.1 70B | 36.5 | #578 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=finreasoning-deep-insight-factuality-and-contextualization · How It Works · Data refreshed daily, snapshot 2026-10-11.