Insight Generation on Internal non-scientific document collections Finance (LLM Judged)
4.56Set Score (Gemini-2.5-Flash)INSIGHTGEN
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| INSIGHTGENBase Model=Claude-3.5-Sonnet2026.04 | 4.56 | 3.33 | |
| INSIGHTGENBase Model=GPT-4o2026.04 | 4.44 | 3.11 | |
| GPT+CoTBase Model=Claude-3.5-Sonnet2026.04 | 4.33 | 3.11 | |
| FAISS+CoTBase Model=GPT-4o2026.04 | 3.67 | 3 | |
| FAISS+CoTBase Model=Claude-3.5-Sonnet2026.04 | 3.44 | 2.89 | |
| GPT+CoTBase Model=GPT-4o2026.04 | 3.33 | 2.67 | |
| FAISSBase Model=GPT-4o2026.04 | 3.22 | 2.11 | |
| Direct GPTBase Model=GPT-4o2026.04 | 3.11 | 2.11 | |
| Direct GPTBase Model=Claude-3.5-Sonnet2026.04 | 2.44 | 1.78 | |
| FAISSBase Model=Claude-3.5-Sonnet2026.04 | 1.67 | 1.67 |