Skip to main content
Eigent supports multiple provider types so open-source deployments can choose models based on capability, cost, privacy, and infrastructure. Provider availability and required fields are defined by the current application. Use this page as an overview and the provider’s official documentation for account, model, and billing details.

​
Cloud and BYOK providers

ProviderTypical required valuesNotes
Google GeminiAPI key, endpoint, modelGoogle model family
OpenAIAPI key, endpoint, modelOpenAI API
AnthropicAPI key, endpoint, modelClaude model family
OrcaRouterAPI key, endpointCan load a grouped model catalog
OpenRouterAPI key, endpoint, modelRoutes models from multiple providers
QwenAPI key, endpoint, modelTongyi Qianwen provider
DeepSeekAPI key, endpoint, modelDeepSeek model family
MiniMaxAPI key, endpoint, modelMiniMax model family
Z.aiAPI key, endpoint, modelZ.ai model family
MoonshotAPI key, endpoint, modelMoonshot model family
ModelArkAPI key, endpoint, modelModelArk service
SambaNovaAPI key, endpoint, modelSambaNova hosted inference
GrokAPI key, endpoint, modelxAI model family
MistralAPI key, endpoint, modelMistral model family
AWS BedrockRegion, access key, secret, modelOptional session token
AWS Bedrock ConverseRegion, access key, secret, modelUses Bedrock Converse integration
Microsoft AzureAPI key, endpoint, API version, deploymentDeployment name is required
Baidu ERNIEAPI key, endpoint, modelERNIE model family
OpenAI-compatibleEndpoint, optional key, modelFor compatible third-party services
Provider fields can change. Confirm required values in Agents > Models after updating Eigent.

​
Local runtimes

RuntimeDefault endpointModel discovery
Ollamahttp://localhost:11434/v1Reads the Ollama tags API
vLLMhttp://localhost:8000/v1Enter the served model when not listed
SGLanghttp://localhost:30000/v1Enter the served model when not listed
LM Studiohttp://localhost:1234/v1Enter the loaded model when not listed
LLaMA.cpphttp://localhost:8080/v1Reads the OpenAI-compatible models endpoint

​
Configure a cloud provider

  1. Create an account with the provider.
  2. Create a restricted API credential.
  3. Confirm the provider endpoint and model identifier.
  4. In Eigent, open Agents > Models.
  5. Select the provider and enter the values.
  6. Validate and save.
  7. Run a small test task.

​
Configure an OpenAI-compatible endpoint

Use the OpenAI-compatible provider for services that implement compatible chat APIs. Provide:
  • Base endpoint
  • API key when required
  • Exact model identifier exposed by the service
Compatibility can vary. Test streaming, tool calls, and structured responses before using the provider for production work.

​
Configure a local runtime

  1. Install and start the runtime.
  2. Load or serve a model.
  3. Confirm the endpoint responds locally.
  4. In Eigent, open Agents > Models > Local.
  5. Select the runtime.
  6. Enter the endpoint and model.
  7. Validate and enable it.
Screenshot placeholder: Add a composite screenshot showing one cloud provider form, Azure provider-specific fields, and one local runtime form. Blur credentials. Video placeholder: Add a 90-second MP4 showing an Ollama model and an OpenAI-compatible cloud endpoint being configured and tested. Include captions.

​
Provider page checklist

Each dedicated provider guide should include:
  1. Account or runtime prerequisites
  2. Credential creation
  3. Endpoint format
  4. Model identifier examples
  5. Eigent configuration
  6. Validation
  7. Common errors
  8. Billing and privacy notes

​
Security guidance

  • Do not include credentials in screenshots, logs, or issue reports.
  • Use least-privilege cloud credentials.
  • Restrict local endpoints to trusted networks.
  • Rotate keys after accidental exposure.
  • Review the provider’s data-retention policy.