GPT-6 Sol Plays Factorio — 100 Green Science | Xhigh

One AI agent. A fresh Factorio world. The challenge: machine-produce 100 green science packs. Watch GPT-6 Sol tackle a FactoryArena benchmark through a Python game interface. This is the full recording, not a highlights edit. RESULT Success: 100/100 target packs produced by team machines. Game time: 0:53:18 Wall-clock benchmark time: 0:53:21 RUN SETTINGS Model: GPT-6 Sol Reasoning effort: Xhigh Game: Factorio 2.0.77 Time limit: None Harness: Codex via ACP Map seed: 19770917 Fresh agent conversation and world; no carried learning. HOW IT WORKS The agent chooses its own construction and logistics strategy. Manual item transfers are allowed; machine-produced science does not imply a fully automated factory. Belts and inserters are optional. This run uses updated instructions explaining automation and production-capacity tradeoffs, without prescribing a build order. Programs use blocking execution; the background-execution comparison is a separate run. MODEL COST Verified API-equivalent cost estimate: not yet available for this session. This run used a Codex subscription; no per-run API invoice is claimed. Hosting, recording and thumbnail generation are excluded. Model costs from other runs are not substituted for this model’s pricing. ABOUT THE BENCHMARK FactoryArena explores how AI agents plan, build and recover from mistakes. The server independently checks the objective. This is a custom scenario, not a default-settings speedrun. Instructions and interface versions may differ between runs. Thumbnail: AI-generated illustrative cover art made with Codex, not an in-game screenshot. #Factorio #AI #FactoryArena