Online Mind2Web
Online Mind2Web is the live, online version of Mind2Web. It does not rely on cached pages and allows for real-time testing against dynamic, evolving web interfaces.
Paper:
An Illusion of Progress? Assessing the Current State of Web Agents (Xue et al., 2025)
Key Features of Online Mind2Web
Real-World Tasks
All tasks are sourced from 136 popular websites to reflect authentic user workflows.
Live Web Evaluation
AI agents interact with live, changing webpages rather than cached static snapshots.
Diverse Tasks
Tasks cover a wide range of scenarios, from clothing and food to housing and transportation.
Online Mind2Web Leaderboard
| Rank | Scaffold |
Primary Model
Primary Model
This is the primary model used by the agent. In some cases, an embedding model is used for RAG, or a secondary model like GPT-4o for image processing. Note: For non-OpenAI reasoning models, the reasoning token budget is set at 1,024 (low), 2,048 (medium), and 4,096 (high). |
Verified
Verified Results
Results have been reproduced by the HAL team |
Accuracy
Accuracy
Confidence intervals show the min-max values across runs for those agents where multiple runs are available |
Cost (USD)
Total Cost
Total API cost for running the agent on all tasks. Confidence intervals show the min-max values across runs for those agents where multiple runs are available |
Runs
Number of Runs
The number of runs for this agent submitted to the leaderboard. To submit multiple evaluations, rerun the same agent and set the same agent name |
Traces |
|---|---|---|---|---|---|---|---|
| 1 |
SeeAct
Pareto optimal
|
GPT-5 Medium (August 2025) | ✓ | 42.33% | $171.07 | 1 | Download |
| 2 | Claude Sonnet 4 (May 2025) | ✓ | 40.00% | $1577.26 | 1 | Download | |
| 3 | Claude-3.7 Sonnet High (February 2025) | ✓ | 39.33% | $1151.88 | 1 | Download | |
| 4 | Claude Sonnet 4 High (May 2025) | ✓ | 39.33% | $1609.92 | 1 | Download | |
| 5 | o3 Medium (April 2025) | ✓ | 39.00% | $258.74 | 1 | Download | |
| 6 | Claude-3.7 Sonnet (February 2025) | ✓ | 38.33% | $926.48 | 1 | Download | |
| 7 | Claude Sonnet 4 (May 2025) | ✓ | 36.67% | $246.18 | 1 | Download | |
| 8 | Claude Sonnet 4 High (May 2025) | ✓ | 36.67% | $326.41 | 1 | Download | |
| 9 | GPT-4.1 (April 2025) | ✓ | 36.33% | $236.62 | 1 | Download | |
| 10 | DeepSeek V3 (March 2025) | ✓ | 32.33% | $214.74 | 1 | Download | |
| 11 | o4-mini High (April 2025) | ✓ | 32.00% | $228.98 | 1 | Download | |
| 12 | GPT-5 Medium (August 2025) | ✓ | 32.00% | $736.31 | 1 | Download | |
| 13 | o4-mini Low (April 2025) | ✓ | 31.67% | $162.36 | 1 | Download | |
| 14 | GPT-4.1 (April 2025) | ✓ | 30.33% | $271.24 | 1 | Download | |
| 15 | Claude-3.7 Sonnet High (February 2025) | ✓ | 30.33% | $367.51 | 1 | Download | |
| 16 |
Browser-Use
Pareto optimal
|
Gemini 2.0 Flash (February 2025) | ✓ | 29.00% | $8.83 | 1 | Download |
| 17 | o3 Medium (April 2025) | ✓ | 29.00% | $371.59 | 1 | Download | |
| 18 | Claude-3.7 Sonnet (February 2025) | ✓ | 28.33% | $291.97 | 1 | Download | |
| 19 |
SeeAct
Pareto optimal
|
Gemini 2.0 Flash (February 2025) | ✓ | 26.67% | $5.03 | 1 | Download |
| 20 | DeepSeek R1 (January 2025) | ✓ | 25.33% | $280.93 | 1 | Download | |
| 21 | o4-mini High (April 2025) | ✓ | 20.00% | $297.93 | 1 | Download | |
| 22 | o4-mini Low (April 2025) | ✓ | 18.33% | $201.44 | 1 | Download |
Accuracy vs. Cost Frontier for Online Mind2Web
This plot shows the relationship between an agent's performance and its token cost. The Pareto frontier (dashed line) represents the current state-of-the-art trade-off. The error bars indicate min-max values across runs.
Heatmap for Online Mind2Web
The heatmap visualizes success rates across tasks and agents. Colorscale shows the fraction of times a task was solved across reruns of the same agent. The "any agent" performance indicates the level of saturation of the benchmark and gives a sense of overall progress.
Total Completion Tokens Used per Agent
The bar chart shows the total completion tokens used by each agent, with the height of each bar representing the total number of completion tokens used across all tasks. Secondary models usually contribute a relatively small amount of tokens in comparison, and are used for RAG or image processing only.
Model Performance Over Time
Track how model accuracy has evolved over time since their release dates. Each point represents the best performance achieved by that model on Online Mind2Web.
Token Pricing Configuration
Adjust token prices to see how they affect the total cost calculations in the leaderboard and plots.
Additional Resources
Getting Started
Want to evaluate your agent on Online Mind2Web? Follow our comprehensive guide to get started:
View DocumentationTask Details
Browse the complete list of Online Mind2Web tasks, including problem descriptions and test cases:
View Tasks