GAIA - General AI Assistant
Multi-agent multi-modal multi-model AI platform with high agency, including browser automation, code generation & execution, and multi-modal reasoning. It can solve many GAIA Benchmark level 1, 2, and 3 problems. To get started with the GUI, select from the examples below. To use via API or MCP, see the link below. Bring your own API keys.
GAIA Benchmark Level 1 Problems
| Question * | GAIA Benchmark Level | Ground Truth | File Name |
|---|
Pages:
GAIA Benchmark Level 2 Problems
| Question * | GAIA Benchmark Level | Ground Truth | File Name |
|---|
Pages:
...
GAIA Benchmark Level 3 Problems
| Question * | GAIA Benchmark Level | Ground Truth | File Name |
|---|
Pages:
Example of GAIA successfully answering a GAIA Benchmark level 3 question:
- GUI (Gradio)
- API (Postman HTTP client calling Gradio REST API #1)
- API (Postman HTTP client calling Gradio REST API #2)
- MCP (Postman MCP client calling Gradio MCP server)
Built with Gradio, crewAI & Arize AI using OpenAI, Gemini & Anthropic models. Tested with Postman (filter by GAIA). By Bernd Straehle.