List your company

Retrieval-Augmented Question Answering on Unique Canonical User-Facing Queries n=170 (Overall)

97FaithfulnessClaude Opus 4.7
16.9237.7158.579.29Jun 14, 2026

Evaluation results

MethodLinks
Claude Opus 4.72026.0697—4533
Naive RAG2026.0687727277
I-GUIDE SSmode=Smart Search2026.0681989197
Naive RAG2026.0678384141
I-GUIDE SSmode=Smart Search2026.0677737271
Claude Opus 4.72026.0676—5650
Qwen2.52026.0669—3951
GPT-5.52026.0664—5572
Qwen2.52026.0662—4157
GPT-4o2026.0658—4461
GPT-4o2026.0620—2782
GPT-5.52026.0620—2685