Orak - Darkest Dungeon: leaderboard
Orak Darkest Dungeon: an LLM agent leads a party through the first expedition; score combines combats cleared (40 points), heroes surviving (30) and remaining stress (30), 0–100, averaged over trials with the game's default agentic strategy. Higher is better. From KRAFTON AI's Orak benchmark (NeurIPS 2025).
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) | 93.7 | #126 |
| 2 | GPT-4o (2024-08-06) | 93.4 | #343 |
| 3 | DeepSeek R1 | 91.7 | #262 |
| 4 | Claude 3.7 Sonnet | 89.9 | #259 |
| 5 | O3 Mini (2025-01-31) | 89 | #249 |
| 6 | Qwen 2.5 7B Instruct | 88.8 | #863 |
| 7 | GPT-4o Mini | 81.3 | #598 |
| 8 | Llama 3.2 3B Instruct | 47.5 | #1327 |
| 9 | Qwen 2.5 3B Instruct | 44.8 | #1159 |
| 10 | Llama 3.2 1B Instruct | 0 | #1518 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=orak-darkest-dungeon · How It Works · Data refreshed daily, snapshot 2026-11-05.