Kaggle Adversarial Customer Service (Gert Labs): leaderboard
Gert Labs' Adversarial Customer Service benchmark on Kaggle (leaderboard version 1): LLM agents compete in a hidden-information customer service simulation conducted in natural language. Score is the task result that Kaggle displays as a percentage; higher is better, 0-100.
Metric: Score (%). Source: www.kaggle.com. Saturation forecast: Not forecast. 11 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.8 | 78.33 | #39 |
| 2 | Gemini 3.6 Flash | 78.33 | #57 |
| 3 | Claude Sonnet 5.5 | 77.5 | #14 |
| 4 | Gemini 3.5 Flash | 76.67 | #66 |
| 5 | Claude Opus 5 | 75.83 | #16 |
| 6 | GPT-5.6 Sol | 75 | #22 |
| 7 | GPT-6 Astra | 75 | #6 |
| 8 | GPT-5.6 Luna | 75 | #79 |
| 9 | GPT-5.6 Terra | 74.17 | #40 |
| 10 | Gemini 3.7 Flash | 73.33 | #24 |
| 11 | Gemini 3.5 Flash Lite | 63.33 | #174 |
No result here: #3 Claude Opus 5.5, #7 GPT-6.1 Sol, #9 Gemini 4 Argon.
Interactive version: theaggregate.ai/benchmark?slug=kaggle-adversarial-customer-service-gert-labs · How It Works · Data refreshed daily, snapshot 2026-11-05.