RPCBench - Domain - Goodreads Dual-Domain - 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 | 1.82 |
| 2 | Qwen 3.5 Plus | 2.05 |
| 3 | Qwen 3.5 35B A3B | 2.28 |
| 4 | GPT-5.5 | 3.19 |
| 5 | Qwen 3.5 397B A17B | 3.19 |
| 6 | Gemini 3.1 Pro (Preview) | 9 |
| 7 | DeepSeek V4 Pro | 9.11 |
| 8 | DeepSeek V4 Flash | 10.36 |
| 9 | Claude Sonnet 4.6 | 17.08 |
| 10 | Llama 3.1 70B Instruct | 44.31 |
| 11 | Llama 3.1 8B Instruct | 48.86 |
Interactive version: theaggregate.ai/benchmark?slug=rpcbench-domain-goodreads-dual-domain-fact-fabrication · How It Works · Data refreshed daily, snapshot 2026-09-24.