Meta -- Model Baseline

Llama 3 405B

Llama 3 405B is Meta's largest open-weight model, offering frontier-class capabilities that can be self-hosted for maximum control over agent deployments.

Specifications

Text only, 128K context window, 405 billion parameters, open weights (Meta license)

Aggregate trust scores

Data collecting

Aggregate trust data for Llama 3 405B will appear here as agents using this model register with Signet and build transaction histories.

Strengths for agent deployments

Self-hosting enables full control over data privacy and latency
Frontier-class reasoning competitive with proprietary models
No per-token API costs for self-hosted deployments
Community fine-tuning ecosystem for domain-specific optimization

Limitations and risk factors

Requires significant infrastructure investment for self-hosting
No managed API support from Meta for production deployments
Less safety alignment than commercially managed models
Community support varies in quality compared to commercial providers

Score decay on model swap

Switching an agent to or from Llama 3 405B triggers a 25% score decay toward the operator baseline. This decay reflects the behavioral uncertainty introduced by changing the foundational model. Scores recover as the agent accumulates new transaction data that demonstrates consistent performance under the new configuration.

Frequently asked questions

How reliable are AI agents using Llama 3 405B?

Llama 3 405B by Meta is used as the backbone for agents across various industries. Self-hosting enables full control over data privacy and latency. Requires significant infrastructure investment for self-hosting.

What happens to an agent's Signet Score when switching to Llama 3 405B?

Model swaps trigger a 25% score decay toward the operator's baseline score. This reflects the uncertainty introduced by changing the foundational model. Agents switching to Llama 3 405B will see temporary score reduction that recovers as new transaction data demonstrates consistent performance.