v0.1.0-beta.1 — Initial Public Beta

Overview

This is the initial public beta of Ego Browser Benchmark Framework, an open-source and reproducible harness for comparing model browser-task performance under controlled agent and browser-tool settings.

Highlights

• Compares models while holding the provider route, browser runtime and skill, task set, judge, limits, and concurrency fixed.
• Supports Pi and Codex as agent providers.
• Provides independent run, judge, and report stages.
• Includes a local WebUI for configuring, monitoring, rejudging, and exporting runs.
• Records evidence and session artifacts for auditable comparisons.

Beta status

Interfaces, configuration, dataset snapshots, and report schemas may change before 1.0. Results are intended for controlled comparisons, not as a universal leaderboard.

Verification

• 296 automated tests passed.
• Ruff checks passed.
• Frontend production build passed.
• GitHub CI passed for Python and frontend jobs.