Orak - Her Story: leaderboard
Orak Her Story: an LLM agent searches a database of police interview clips by keyword to uncover the hidden story; score is based on the number of distinct clips needed to complete the game, normalized to 0–100 and averaged over trials with the default agentic strategy. Higher is better. From KRAFTON AI's Orak benchmark.
Metric: Game Score (0-100). Source: krafton-ai.github.io. Saturation forecast: Estimated already saturated. 12 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro (Preview 03-25) | 67.3 | #126 |
| 2 | DeepSeek R1 | 66.9 | #262 |
| 3 | O3 Mini (2025-01-31) | 66.3 | #249 |
| 4 | GPT-4o (2024-08-06) | 64 | #343 |
| 5 | Claude 3.7 Sonnet | 62.6 | #259 |
| 6 | GPT-4o Mini | 21.1 | #598 |
| 7 | Qwen 2.5 7B Instruct | 8.5 | #863 |
| 8 | Llama 3.2 3B Instruct | 4.2 | #1327 |
| 9 | Llama 3.2 1B Instruct | 2.1 | #1518 |
| 10 | Qwen 2.5 3B Instruct | 1.2 | #1159 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=orak-her-story · How It Works · Data refreshed daily, snapshot 2026-11-05.