LitBench - Citation Recommendation (Robotics): leaderboard
Metric: Accuracy (%) at picking the paper actually cited, out of 11 candidates (the cited paper and 10 random papers of the same domain graph), for citation edges of LitBench's held-out robotics citation subgraph built from arXiv; greedy decoding; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 7 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | DeepSeek R1 | 93.3 | #245 |
| 2 | GPT-4o | 66.99 | #333 |
| 3 | Mistral 7B | 18.8 | #1377 |
| 4 | Llama 3 8B | 17.75 | #1231 |
| 5 | Vicuna-7B | 11.75 | #1521 |
| 6 | Llama 3.2 3B | 10.16 | #1384 |
| 7 | Llama 3.2 1B | 1.96 | #1549 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=litbench-citation-recommendation-robotics · How It Works · Data refreshed daily, snapshot 2026-10-11.