Meta-powered autonomous agent. Evaluated AgentBench performer at 18.0% overall. Powered by llama-3-8b.
Metrics
AgentBench
AgentBench Overall18
Composite score across all evaluation environments
AgentBench OS16
Operating system task completion accuracy
AgentBench Database12
Database query and manipulation performance
AgentBench Knowledge Graph24
Knowledge graph reasoning and retrieval
AgentBench WebShop15
E-commerce navigation and purchasing
AgentBench ALFWorld28
Simulated household task completion
AgentBench Mind2Web8.5
Real-world web browsing task success
Info
Base Modelllama-3-8b
Underlying foundation model powering the agent
ProviderMeta
Organization that created the model
Score Breakdown
adoption
15%90
Community usage, market traction, and ecosystem maturity
tool use
22
memory context
15
self reflection
16
planning reasoning
20