Open Source • Model Agnostic • AI Orchestration

OpenModelFlow

Build AI workflows that work across models, providers and tools.

OpenModelFlow is an open-source AI orchestration framework for building reliable, observable and provider-independent AI applications. Design workflows once and execute them with any supported model without changing your application logic.

Model Agnostic

Run the same workflow across OpenAI, Anthropic, Google, Ollama and other Open Models.

Tool Orchestration

Execute tools, APIs, MCP servers and custom functions with a consistent interface.

Workflow Engine

Compose sequential, parallel, conditional and multi-step AI workflows with ease.

Streaming

Support real-time streaming responses while preserving workflow state.

Observability

Built with execution tracing, retries, execution history and runtime inspection in mind.

Developer First

Simple APIs, extensible architecture and predictable execution designed for production.

Core Principles

Provider Independence

Switch AI providers without rewriting your workflows.

Composable Workflows

Build reusable AI pipelines from simple building blocks.

Reliable Execution

Support retries, error handling and execution state throughout the workflow.

Unified Tool Calling

Use one interface for tools regardless of the underlying model.

Extensible Runtime

Add new providers, tools and execution strategies with minimal effort.

Open by Design

Community-driven, transparent and built around open standards.

Designed For

AI AgentsAgentic WorkflowsEnterprise AutomationTool CallingMCP IntegrationMulti-Model ApplicationsLLM OrchestrationProduction AI SystemsBackend ServicesAI Platforms

One Workflow. Any Model.

Focus on building intelligent applications instead of managing provider-specific APIs, tool formats and orchestration logic. OpenModelFlow provides a unified runtime for modern AI systems.