Papers & Research

EVA-Bench Data 2.0 released: three domains, 121 tools, 213 scenarios

Original title:EVA-Bench Data 2.0: 3 Domains, 121 Tools, 213 Scenarios

AI summary

EVA-Bench Data 2.0 expands evaluation from a single enterprise domain to three: airline customer service management (CSM), enterprise IT service management (ITSM) and healthcare HR service delivery (HRSD), covering 121 tools and 213 scenarios, about 4x the number of scenarios in the original version. Each scenario was verified as solvable by OpenAI GPT-5.4, Google Gemini 3.1 Pro and Anthropic Claude Opus 4.6. The dataset follows five design principles — voice-first, realism, diversity, authentication flows and reproducibility — and includes single-intent, multi-intent (up to 4 intents) and adversarial call types. All three datasets are open-source and can be downloaded directly from Hugging Face via `load_dataset`. A multilingual extension will follow.

Key Takeaways

  • EVA-Bench Data 2.
  • 0 expands evaluation from a single enterprise domain to three: airline customer service management…
  • Each scenario was verified as solvable by OpenAI GPT-5.
  • 4,Google Gemini 3.
  • 1 Pro and Anthropic Claude Opus 4.

Quick Facts

Published
Updated
Source
Hugging Face 社区博客(混合发现)
Category
Papers & Research
AI score
65
Read the original Read the full article at the original source
Why it matters

语音代理评测缺的就是这种真实场景的数据集,EVA-Bench 2.0 把航空、IT、医疗三个最棘手的领域打包了,生成流水线也开源,做评测的可以直接拿来用。

Related entities

Source & copyright

This article is based on curated titles and AI summaries from the AIHOT public API, with attribution and links back. Original copyright remains with the original publishers and authors; this site does not mirror full texts.

Original article · AIHOT entry

How to Cite

When citing this article, keep the title, date, and permalink, and credit Orange News as the source.

Orange News: EVA-Bench Data 2.0 released: three domains, 121 tools, 213 scenarios. June 4, 2026. https://new.nat6.net/en/news/cmpzhk1rp03k5slkpmx3q7mp3/

View all →