A Coding Agent for Local Models
Point Felan's agents at a local or private model endpoint, not only a hosted provider. Switch between them per task, with the same agent and workflow.
Local and Private Models, Not Just Hosted Ones
Alongside Anthropic, OpenAI, Google, DeepSeek, and Moonshot AI, connect a local model or a private endpoint you control.
Hosted
- Anthropic
- OpenAI
- DeepSeek
- Moonshot AI
Yours
- Local model
- Private endpoint
Match the Model to the Task
Use a cheaper or local model for simple work, and a top hosted model for hard problems, and switch whenever prices or quality change.
- Update copy stringsCheaper or local model
- Trace a memory leakTop hosted model
Your Endpoint, Your Control
A private or local model connects the same way any other provider does: a saved key or endpoint, ready for any agent to use.
Private model
ConnectedAvailable to any agent
Common Questions
Can Felan run agents against a local model?
Yes. Connect a local model or your own private endpoint alongside hosted providers such as Anthropic, OpenAI, Google, DeepSeek, and Moonshot AI.
What counts as a private model?
Any model you serve yourself and expose as an endpoint, instead of using a hosted provider's API.
Does Felan mark up local model usage?
No. Felan doesn't mark up tokens on any provider, including local and private endpoints. You bring your own infrastructure and control its cost.
Can I mix local and hosted models?
Yes. Choose a model per task, so simple work can use a local or cheaper model while hard problems use a top hosted model.
Is local model support available on every plan?
Local and private models are available wherever you bring your own keys or endpoint, which every plan supports.
Learn more: Predictable token costs, Bring your own keys.
Keep Your Attention for the Hard Problems
Start with one repo and one recurring task.