Post

The next era of Android Bench! 🧵

8/7/26, 4:11 pm2 reposts12 likes

🔹 8 new models added: Claude Fable 5 (leading at 84.5%), GPT 5.5, Sonnet 5, GLM 5.2, and Qwen 3.7 Max are now on the board.

🔹 Upgraded methodology: Now using the Harbor Framework for tougher, real-world AI reasoning tests.

🔹 Community contributions: Starting today, you can submit your own dev tasks and share evaluations on Harbor Hub.

Check out the updated leaderboard and the full methodology breakdown: goo.gle/4wB7Vqk