Model Providers
The Model Provider configuration is a centralized setup that defines which AI model connections are available across your application. It allows you to configure model providers, select models, and manage authentication details from a single location.
By using a global configuration, multiple agents and workflows can share the same provider settings without requiring individual setup. This ensures consistency, simplifies management, and reduces duplication across the platform.
Supported providers include OpenAI, Anthropic, Google Generative AI, IBM WatsonX, and Ollama. Each provider stores its API key and model details in one place, enabling all model-driven components to access the same connection and credentials seamlessly.
Configuring a Global Model Provider in Robility Flow
Before using AI models in your workflows, you need to connect a model provider and enable the required models.
Step 1: Access Model Provider Settings
1. Click your profile icon in the top corner
2. Navigate to Settings → Model Providers
Step 2: Select a Provider
1. Choose the provider you want to connect (for example, OpenAI or Anthropic)
Step 3: Add API Key
1. Enter your provider’s API key in the API Key field
Ensure that your API key:
a. Has permission to access the required models
b. Is associated with sufficient usage credits
Note: Only one API key can be configured per provider. Choose a key that supports all required models. You can update or replace it at any time.
Step 4: View Available Models
Once the API key is added, available models appear under:
Language Models — used for generating text responses
Embedding Models — used for generating vector embeddings
The available models depend on the provider and the permissions associated with your API key.
Step 5: Use Models in Workflows
1. Configured models become available across all workflows
2. No additional setup is required in individual components
Example
After connecting a provider, it will appear in model selection lists across the platform. For example, in an agent’s Language Model settings, all connected providers and their models are available for selection.
