AppWorld-UL (Function Calling): leaderboard

Metric: Interactive task goal completion (%; share of all 516 user-in-the-loop tasks (306 single-interaction and 210 compositional) whose AppWorld unit tests all pass, including a test that the agent asked the simulated user for every piece of hidden information it needed; function-calling agent over the AppWorld app APIs; GPT-5.5 user simulator answering from a fixed knowledge set). Source: arxiv.org. Saturation forecast: Around 2030. 5 models tracked.

Top models

#ModelScore
1Claude Opus 4.727.3
2Qwen 3.7 Max25.7
3DeepSeek V4 Pro25.6
4Kimi K2.624.4
5GPT-5.5 (Medium)24.2

Interactive version: theaggregate.ai/benchmark?slug=appworld-ul-function-calling · How It Works · Data refreshed daily, snapshot 2026-09-29.