WebArena
by CMU · open-source · Last verified 2026-03-01
Realistic web environment benchmark testing agents' ability to complete complex tasks across self-hosted websites including e-commerce, forums, collaborative development, and content management platforms with real web interfaces.
https://webarena.dev ↗B
B—Above Average
Adoption: BQuality: A+Freshness: ACitations: B+Engagement: F
Specifications
- License
- Apache-2.0
- Pricing
- open-source
- Capabilities
- agent-evaluation, web-interaction-testing, browser-automation-assessment
- Integrations
- playwright, docker
- Use Cases
- web-agent-benchmarking, browser-automation-evaluation, interactive-agent-testing
- API Available
- No
- Evaluated Models
- claude-4, gpt-5, gemini-2.5-pro, deepseek-v3
- Metrics
- success-rate, step-accuracy
- Methodology
- 812 web-based tasks across 5 self-hosted websites. Agents interact via browser actions and are evaluated on task completion determined by URL, page content, or database state checks.
- Last Run
- 2026-02-28
- Tags
- benchmark, evaluation, agents, web, browser-automation
- Added
- 2026-03-17
- Completeness
- 100%
Index Score
62.4Adoption
66
Quality
90
Freshness
86
Citations
72
Engagement
0