RPCBench - Domain - Amazon Sports - Evidence Faithfulness: leaderboard
RPCBench tests proactive premise critique in recommendation requests. This series measures evidence faithfulness on the Amazon Sports subset (1,270 questions), using the same eleven-model, three-judge experiment as the headline paper. Published EFI fractions from the authors' per-model CSV are scaled once to Faithfulness Index (%). Higher is better; domains and nested premise-error groups are related slices of the same experiment.
Metric: Faithfulness Index (%). Source: github.com. Saturation forecast: Estimated already saturated. 11 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.5 | 91.61 | #48 |
| 2 | Qwen 3.5 Plus | 91.3 | #232 |
| 3 | Qwen 3.5 397B A17B | 90.75 | #245 |
| 4 | Qwen 3.5 122B A10B | 90.16 | #335 |
| 5 | Qwen 3.5 35B A3B | 88.94 | #492 |
| 6 | Gemini 3.1 Pro (Preview) | 85.91 | #93 |
| 7 | Claude Sonnet 4.6 | 82.17 | #159 |
| 8 | DeepSeek V4 Pro | 81.18 | #188 |
| 9 | DeepSeek V4 Flash | 80.55 | #259 |
| 10 | Llama 3.1 70B Instruct | 50.24 | #1629 |
| 11 | Llama 3.1 8B Instruct | 44.09 | #4318 |
No result here: #4 Claude Opus 5.5, #5 GPT-6 Astra, #9 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=rpcbench-domain-amazon-sports-evidence-faithfulness · How It Works · Data refreshed daily, snapshot 2026-11-08.