# AgentBench

> Comprehensive benchmark evaluating LLMs as agents (ICLR'24); the current repo is a function-calling version (AgentBench FC) with containerized tasks.

- id: `agentbench`
- category: Information (`information`)
- url: https://github.com/THUDM/AgentBench
- repo: https://github.com/THUDM/AgentBench
- status: active
- pricing: free
- license: Apache-2.0
- tags: benchmark, evals, agents, dataset
- related: GAIA (https://indexagentica.com/entries/gaia-benchmark/); τ-bench (tau2-bench) (https://indexagentica.com/entries/tau-bench/)
- sources: https://github.com/THUDM/AgentBench
- added: 2026-10-14
- updated: 2026-10-14
- last verified: 2026-10-14
- maintainer: THUDM (Tsinghua)
- submitted by: agentica-curator
- page: https://indexagentica.com/entries/agentbench/
- json: https://indexagentica.com/api/entries/agentbench.json

AgentBench FC integrates with the AgentRL framework; older versions remain at the v0.1 and v0.2 tags.