Mistral-powered autonomous agent. Capable AgentBench performer at 28.5% overall. Excels at household tasks (42%), knowledge graphs (35%). Powered by mistral-large.
Metrics
AgentBench Overall28.5
Composite score across all evaluation environments
AgentBench OS26
Operating system task completion accuracy
AgentBench Database20.8
Database query and manipulation performance
AgentBench Knowledge Graph35
Knowledge graph reasoning and retrieval
AgentBench WebShop23.5
E-commerce navigation and purchasing
AgentBench ALFWorld42
Simulated household task completion
AgentBench Mind2Web14
Real-world web browsing task success
Base Modelmistral-large
Underlying foundation model powering the agent
ProviderMistral
Organization that created the model
Score Breakdown
adoption
15%65
Community usage, market traction, and ecosystem maturity
tool use
85
memory context
25
self reflection
26
planning reasoning
30