GuardianAgentBench (LangChain): leaderboard
GuardianAgentBench (Vectara, 2026) runs tool-using agents through 580 business scenarios in six domains, a third of them with adversarially modified tool responses, and requires both the final answer and the tool-call trace to match the ground truth. Every model runs as a LangChain ReAct agent built with LangGraph. Overall correctness in percent averaged over the six domains, judged by Claude Sonnet 4.5; higher is better.
Metric: Overall correctness (%; mean over six agent domains of the share of scenarios whose answer and tool-call trace are both judged correct by Claude Sonnet 4.5; 580 scenarios with 81 tools, a third with adversarially modified tool responses; LangGraph ReAct agent, default API settings). Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 6 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | LangChain + Claude Opus 4.5 | 74 | |
| 2 | LangChain + Gemini 3 Pro | 73.8 | |
| 3 | LangChain + GPT-5.2 Pro | 69.7 | |
| 4 | LangChain + GPT-OSS-120B | 68.3 | |
| 5 | LangChain + Qwen 3 Max | 67.7 | |
| 6 | LangChain + DeepSeek V3.2 | 66.7 |
Interactive version: theaggregate.ai/benchmark?slug=guardianagentbench-langchain · How It Works · Data refreshed daily, snapshot 2026-11-13.