GPT-6 models duel in DOOM: Astra wins 82.5% of 120 agent matches
OpenAIDevs · x · 2026-09-25
Rootly AI Labs open-sourced rootly-doom-agent-arena, an MCP-native DOOM arena where LLM agents fight each other for real-time model-vs-model evaluation.
- 120 matches across GPT-6 Astra, Sol, and Luna (medium reasoning effort), 40 matches each, 20 per pair with side swaps
- Astra topped the leaderboard with an 82.5% win rate (33–7), 83.2% accuracy, +86.25 damage differential — but the highest cost
- Sol decided fastest at an average of 5.34s, with a 42.5% win rate
- Luna won only 25% but offered the best wins-per-dollar (12.8)
All matches ended in elimination with no draws. The arena is available on GitHub for running your own duels.
More from coding & agent
- Prompt boxes are dying, and embedded app agents still lose to agents driving human tools — andpoul · 2026-09-25
- Agents inside apps vs. operating human tools: a short debate — andpoul · 2026-09-25
- Devin launches a redesigned home page for its AI coding agent — DevinAI · 2026-09-25
- Blender agent skill turns concept art into rigged game assets — one user shipped a Mario 64-style game in 8 hours — majidmanzarpour · 2026-09-25
- Designer argues agent-native pro tools, not prompt boxes, are the future of AI work — bartonsmith · 2026-09-25
- Open Manager for ComfyUI adds desktop mode for Vast.ai and other cloud GPU services — WASasquatch · 2026-09-25