Inkling from @thinkymachines on ARC-AGI (Verified)
- ARC-AGI-2: 36.5%, $0.64/task
- ARC-AGI-1: 79.5%, $0.30/task
As of today, Inkling is the highest-scoring open-weight model evaluated by ARC Prize on both ARC-AGI-1 and ARC-AGI-2.
GPT-5.6 Sol just beat Pokémon Emerald. Fully autonomous, vision-only: screenshots in, button presses out
Codex App Server + custom tools + one persistent goal: "Finish the game"
153h51 of active runtime to Champion 🏆