TimeLitmus - Traffic Series-Side Counterfactual Pairs: leaderboard
Metric: Counterfactual pair correctness (%; both independently evaluated endpoints correct on 136 Traffic pairs whose pre-report traffic series is changed while the incident is kept; ten LLMs with deterministic decoding where available, invalid or unparsable outputs scored incorrect). Source: arxiv.org. Saturation forecast: Around 2039. 10 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | DeepSeek R1 0528 | 21.3 |
| 2 | Qwen 3.5 4B | 20.6 |
| 3 | DeepSeek V4 Flash | 18.4 |
| 4 | Qwen 3.5 9B | 16.9 |
| 5 | Qwen Plus | 16.9 |
| 6 | MiniMax-M3 | 11.8 |
| 7 | Claude Sonnet 4.6 | 11 |
| 8 | GPT-5.4 | 8.8 |
| 9 | Gemini 3.5 Flash | 8.8 |
| 10 | GLM-5 | 4.4 |
Interactive version: theaggregate.ai/benchmark?slug=timelitmus-traffic-series-side-counterfactual-pairs · How It Works · Data refreshed daily, snapshot 2026-09-26.