An eval-certified, multi-provider LLM router with automated model lifecycle management.
Model Fleet is an experimental Go project for routing LLM requests across independent providers while preserving application-specific behavioral guarantees.
- Fail over across independent provider quotas.
- Encapsulate provider differences behind stateless clients.
- Certify model deployments for application roles through application-owned evaluations.
- Monitor model availability and deprecations.
- Automate maintenance through bounded, auditable workflows.
Planning and Milestone 1 development.
Licensed under the MIT License.