Fangcun DeepSeek 1017 Report - Lie-to-Me affected-case CoT disclosure: leaderboard
Fangcun report 2610.35671v1, hosted API evaluations Oct 4–25 2026; Appendix A.1 exact model identities, maximum reasoning effort unless named otherwise, temperature 1/top_p 0.95. GLM 5.3 judge where required. API/format failures excluded by source; common cohort versus successfully completed denominators retained. Primary PDF printed values, not estimated plot heights. Figure 1.4b four DeepSeek native CoT models; disclosure of misleading hint among responses whose answer changed to wrong hinted option. Higher disclosure is source-defined transparency quality, not correctness. Conditional affected-case populations differ; GPT/Gemini have no comparable native CoT. Metric: Native rate (%).
Metric: Native rate (%). Source: arxiv.org. Saturation forecast: Estimated already saturated. 4 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | DeepSeek V4.1 Flash (1017) (Max) | 98.3 | #25 (DeepSeek V4.1 Flash (1017)) |
| 2 | DeepSeek V4.1 Flash (Max) | 98.2 | #36 (DeepSeek V4.1 Flash) |
| 3 | DeepSeek V4 Pro (Max) | 98.1 | #109 (DeepSeek V4 Pro) |
| 4 | DeepSeek V4 Flash (Max) | 97.3 | #147 (DeepSeek V4 Flash) |
No result here: #2 Claude Opus 5.5, #3 GPT-6 Astra, #6 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=fangcun-deepseek-1017-report-lie-to-me-affected-case-cot-disclosure · How It Works · Data refreshed daily, snapshot 2026-10-28.