RPCBench - Domain - MovieLens-1M - Fact Fabrication: leaderboard
Metric: Fact Fabrication Rate (%). Source: github.com. Saturation forecast: Estimated already saturated. 11 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3.5 122B A10B | 4.55 |
| 2 | GPT-5.5 | 6.06 |
| 3 | Qwen 3.5 Plus | 6.49 |
| 4 | Qwen 3.5 397B A17B | 7.36 |
| 5 | Qwen 3.5 35B A3B | 7.58 |
| 6 | Gemini 3.1 Pro (Preview) | 7.79 |
| 7 | DeepSeek V4 Pro | 11.26 |
| 8 | DeepSeek V4 Flash | 16.67 |
| 9 | Claude Sonnet 4.6 | 27.71 |
| 10 | Llama 3.1 70B Instruct | 57.14 |
| 11 | Llama 3.1 8B Instruct | 67.53 |
Interactive version: theaggregate.ai/benchmark?slug=rpcbench-domain-movielens-1m-fact-fabrication · How It Works · Data refreshed daily, snapshot 2026-09-24.