T-Eval

A bilingual tool-use benchmark that scores instruction, planning, reasoning, retrieval, understanding and review as separate steps rather than one success bit.

Also known as: TEval, Tool-Eval

unassessed

This page is a discovery lead. Nobody has yet assessed it against the catalogue contract, so it carries no disposition. Absence of evidence here is not evidence of staleness.
Categoryagentic
Subcategorystep-wise LLM tool utilization
Page statusactive
Metricoverall (mean of subset metrics)
Directionhigher_is_better
Dataset size23305
Dataset licenceApache-2.0
PublisherUniversity of Science and Technology of China / Shanghai AI Laboratory / Tsinghua University / Jilin University / OpenCompass

What it measures

T-Eval tests whether a model can use tools by grading each sub-skill on its own. Given a query and a tool list, the model must plan, pick tools, fill parameters, emit a call, and review the tool response. The authors argue that scoring only the final answer hides which of those steps failed. English and Chinese prompts both exist. This is not a coding benchmark and not ToolBench's end-to-end win rate, though the paper plots the two against each other.

Task format

Multi-turn chat with system/user/assistant roles and a tool schema in the prompt. OpenCompass uses `ChatInferencer` and `TEvalEvaluator` on local JSON files named `instruct_v1`, `plan_json_v1`, `plan_str_v1`, and similar. Hugging Face `lovesnowbest/T-Eval` currently ships v2 files (`instruct_v2.json`, ...).

Models reporting this benchmark

No model card in ModelSpec reports this benchmark yet.

Data

This page as JSON · Edit on GitHub