Drift-Bench (Service-Oriented, No Clarification) - Parameter: leaderboard
Metric: Success rate (%; 150 StableToolBench G1 API tasks with a parameter fault (a wrong or missing value), no clarification tools, mean of 3 trials). Source: arxiv.org. Saturation forecast: Around 2029. 7 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3 235B A22B 2507 Instruct | 67 |
| 2 | DeepSeek V3.2 | 64.11 |
| 3 | Llama 4 Maverick | 57.22 |
| 4 | GLM-4.7 | 57.22 |
| 5 | Gemini 2.5 Flash | 49.67 |
| 6 | GPT-5.2 | 46.56 |
| 7 | GPT-OSS-120B | 40.78 |
Interactive version: theaggregate.ai/benchmark?slug=drift-bench-service-oriented-no-clarification-parameter · How It Works · Data refreshed daily, snapshot 2026-09-25.