AuthorizationBench - OpenAI SDK - Safe Completion Rate: leaderboard
AuthorizationBench's safe completion rate for OpenAI Agents SDK 0.20.0 on 286 executable business tasks. Each model has separate low/high reasoning runs under a 300-second, 35-turn budget; an in-flight turn may finish later. Safe completion requires finishing without acting outside user authorization. All assigned trajectories remain in the denominator; MiMo v2.6 Pro supplies oracle-user replies and Claude Opus 5.5 judges outcomes. Source API/harness/effort identities are preserved.
Metric: Safe Completion Rate (%). Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 8 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.6 Sol (High) (AuthorizationBench Run: OpenAI Agents SDK 0.20.0; 300s; 35 turns) | 86.4 | |
| 2 | Claude Sonnet 5 (Low) (AuthorizationBench Run: OpenAI Agents SDK 0.20.0; 300s; 35 turns) | 80.4 | |
| 3 | GPT-5.6 Sol (Low) (AuthorizationBench Run: OpenAI Agents SDK 0.20.0; 300s; 35 turns) | 80.1 | |
| 4 | Claude Sonnet 5 (High) (AuthorizationBench Run: OpenAI Agents SDK 0.20.0; 300s; 35 turns) | 75.5 | |
| 5 | GLM-5.3 (Low) (AuthorizationBench Run: OpenAI Agents SDK 0.20.0; 300s; 35 turns) | 69.9 | |
| 6 | Gemini 3.8 Flash (Low) (AuthorizationBench Run: OpenAI Agents SDK 0.20.0; 300s; 35 turns) | 68.9 | |
| 7 | GLM-5.3 (High) (AuthorizationBench Run: OpenAI Agents SDK 0.20.0; 300s; 35 turns) | 68.9 | |
| 8 | Gemini 3.8 Flash (High) (AuthorizationBench Run: OpenAI Agents SDK 0.20.0; 300s; 35 turns) | 66.8 |
Interactive version: theaggregate.ai/benchmark?slug=authorizationbench-openai-sdk-safe-completion-rate · How It Works · Data refreshed daily, snapshot 2026-11-05.