Heartbeat badge

Research Platform for Optimizing
Conversational AI Apps

a benchmark for emotional intelligence | derived from EQ-Bench 4

Application menu

Dashboard for starting a new EQ evaluation Evaluation results for a previous EQ evaluation Evaluation results — comprehensive report Track cost and latency metrics Search the database Query the database

List Rubrics Manage a Rubric List Criteria Manage a Criterion
List Scenario Collections Manage a Scenario Collection List Scenarios Manage a Scenario List User Personas Manage a User Persona

List Eval Prompts View a selected Eval Prompt List Eval Judge Results View a selected Eval Judge Result List Eval Conversations View a selected Eval Conversation

Log out Sign up — Create a new member account Manage my account Reset my password

Welcome to this research prototype derived from EQ-Bench 4!

Administrator reports Administrator tools Administrator tools

()

My account profile logoutIcon

Evaluation Metrics

Select a results JSON file to view: (default location: eqbench4/results/)

Summary

--

Aggregated Score
(weighted average)

--

User State (50%)

--

EQ Rubric (25%)

--

Debrief (25%)

--

Actual Cost

Score Breakdown

Cost Tracking

Total Cost: --

--

Conversations

Run Selection

Select Run:

Run Summary

Run Metadata

Dataset Information

Aggregate Statistics

Search & Filter

Search All Fields:

Category:

Subcategory:

Core Issue:

Trust Range: -

Progress Range: -

Scenarios

No scenarios match your search/filter criteria.

Compare Conversations

Conversation A

Conversation B

Select two conversations to compare.

Measure emotional intelligence in language-model behaviors
EQ-Bench developed by: Sam Paech (MIT license) | User interface and relational data model developed by: Daniel Webb © 2025-2026