ArchBench Serverless Component Generation - Function Tests: leaderboard
ArchBench serverless function generation scored by the percentage of the generated function's own test cases that pass without human edits, using the paper's most detailed prompt; higher is better (IIIT Hyderabad, arXiv 2502.02539, 2025).
Metric: Function Test Pass Rate (%, no human intervention). Source: github.com. Saturation forecast: Around May 2028. 5 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4 | 10 | #996 |
| 2 | CodeQwen1.5-7B Chat | 7 | #5781 |
| 3 | GPT-3.5 Turbo | 4 | #2773 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=archbench-serverless-component-generation-function-tests · How It Works · Data refreshed daily, snapshot 2026-11-12.