OpenAI models
Use the Responses API through a server-side key and select the model for your workload.
Configure OpenAISelf-hosted agent runtime
Run AI agentson your infrastructure.
Connect OpenAI or your own Ollama server, run task-focused agents, and understand every token from one secure workspace.
Server-side credentials Go runtime Usage insights
openagent.runtimelive
01prompt.receivedready
02ollama.generaterunning
03usage.track184 tokens
Model freedom
One workflow. Your choice of intelligence.
Move between hosted and self-hosted models without changing how your agents work.
Ollama server
Keep inference close to your data with an Ollama endpoint and locally managed models.
Connect OllamaUsage dashboard
Compare requests, input tokens, output tokens, and recent activity by provider.
View agent workspaceInspectable by design
A compact stack you can understand and control.
Open source, straightforward infrastructure, and provider credentials that remain on your server.
Browse the sourceApplicationGo service
InterfaceReact
DataPostgreSQL
Agent accessAuthenticated
Ask Emerald OpenAgent guide