AuthMem-Bench (Task Success Rate): leaderboard
Metric: Task success rate (%; share of 350 authorized counterpart cases in which the action model issues the required tool call; washed focal memory with no authority metadata). Source: arxiv.org. Saturation forecast: Around 2029. 7 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.4 Mini | 56.6 |
| 2 | Gemini 3.1 Pro (Preview) | 52.9 |
| 3 | GPT-5.5 | 52 |
| 4 | Gemini 3.5 Flash | 51.7 |
| 5 | GLM-5.2 | 50.9 |
| 6 | Qwen 3.7 Max | 46 |
| 7 | DeepSeek V4 Pro | 35.7 |
Interactive version: theaggregate.ai/benchmark?slug=authmem-bench-task-success-rate · How It Works · Data refreshed daily, snapshot 2026-09-26.