SynthAPI Paper - Holdout Shopping: leaderboard
SynthAPI tests whether models can use unfamiliar APIs and complete tasks with tools, including AppWorld, ASTRA and function calling. Episode success (%).
Metric: Episode success (%). Source: arxiv.org. Saturation forecast: Around September 2028. 14 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3.5 27B | 8 | #186 |
| 2 | Qwen 3.5 9B | 1 | #387 |
| 3 | Qwen 3.5 4B | 0 | #496 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=synthapi-paper-holdout-shopping · How It Works · Data refreshed daily, snapshot 2026-10-29.