GenAI Processors Library 📚

Build Modular, Asynchronous, and Composable AI Pipelines for Generative AI.

GenAI Processors is a lightweight Python library that enables efficient, parallel content processing. It addresses the fragmentation of LLM APIs through three core pillars:

1. Unified Content Model: A single, consistent representation for inputs and outputs across models, agents, and tools.
2. Processors: Simple, composable Python classes that transform content streams using native asyncio.
3. Streaming: Asynchronous streaming capabilities built-in by default, without added plumbing complexity.

At the ecosystem's core lies the Processor, which encapsulates a unit of work. Through a "dual-interface" pattern, it handles the complexity of asynchronous, multimodal data streaming while exposing a simple API to developers:

from typing import AsyncIterable
from genai_processors import content_api
from genai_processors import processor

class EchoProcessor(processor.Processor):
  # The PRODUCER interface (for the processor author):
  # Takes a robust ProcessorStream as input, and yields part types.
  async def call(
      self, content: content_api.ProcessorStream
  ) -> AsyncIterable[content_api.ProcessorPartTypes]:
      # Process content as it streams in!
      async for part in content:
          yield part

Applying a Processor is just as straightforward. The CONSUMER interface accepts wide, forgiving input types and returns a powerful stream that can be awaited entirely or streamed chunk-by-chunk:

# The CONSUMER interface (for the caller):
# Provide input effortlessly. Strings are automatically cast into Parts.
input_content = ["Hello ", content_api.ProcessorPart("World")]

# 1. Gather all outputs easily into one object:
result: content_api.ProcessorContent = await simple_text_processor(input_content).gather()

# 2. Or for text-only agents, get the text directly:
print(await simple_text_processor(input_content).text())

# 3. And for streaming use cases, iterate over the parts as they arrive:
async for part in simple_text_processor(input_content):
  print(part.text, end="")

The concept of Processor provides a common abstraction for Gemini model calls and increasingly complex behaviors built around them, accommodating both turn-based interactions and live streaming.

✨ Key Features

• Modular: Breaks down complex tasks into reusable Processor and PartProcessor units, which are easily chained (+) or parallelized (//) to create sophisticated data flows and agentic behaviors.
• Integrated with GenAI API: Includes ready-to-use processors like GenaiModel for turn-based API calls and LiveProcessor for real-time streaming interactions.
• Extensible: Lets you create custom processors by inheriting from base classes or using simple function decorators.
• Rich Content Handling:
  • ProcessorPart: A wrapper around genai.types.Part enriched with metadata like MIME type, role, and custom attributes.
  • Supports various content types (text, images, audio, custom JSON).
• Asynchronous & Concurrent: Built on Python's familiar asyncio framework to orchestrate concurrent tasks (including network I/O and communication with compute-heavy subthreads).
• Stream Management: Has utilities for splitting, concatenating, and merging asynchronous streams of ProcessorParts.

📦 Installation

The GenAI Processors library requires Python 3.10+.

Install it with:

pip install genai-processors

Code generation

Generative models are often unaware of recent API and SDK updates and may suggest outdated or legacy code.

We recommend using our Code Generation instructions when generating code that uses GenAI Processors to guide your model towards using the more recent SDK features. Copy and paste the instructions into your development environment to provide the model with the necessary context.

🚀 Getting Started

We recommend to start with the documentation microsite which covers the core concepts, development guides, and architecture.

You can also check the following colabs to get familiar with GenAI processors (we recommend following them in order):

• Content API Colab - explains the basics of ProcessorPart, ProcessorContent, and how to create them.
• Processor Intro Colab - an introduction to the core concepts of GenAI Processors.
• Create Your Own Processor - a walkthrough of the typical steps to create a Processor or a PartProcessor.
• Work with the Live API - a couple of examples of real-time processors built from the Gemini Live API using the LiveProcessor class.

📖 Examples

Explore the examples/ directory for practical demonstrations:

• Real-Time Live Example - an Audio-in Audio-out Live agent with google search as a tool. It is a client-side implementation of a Live processor (built with text-based Gemini API models) that demonstrates the streaming and orchestration capabilities of GenAI Processors.
• Research Agent Example - a research agent built with Processors, comprising 3 sub-processors, chaining, creating ProcessorParts, etc.
• Live Commentary Example - a description of a live commentary agent built with the Gemini Live API, composed of two agents: one for event detection and one for managing the conversation.

🧩 Built-in Processors

The core/ directory contains a set of basic processors that you can leverage in your own applications. It includes the generic building blocks needed for most real-time applications and will evolve over time to include more core components.

Community contributions expanding the set of built-in processors are located under contrib/ - see the section below on how to add code to the GenAI Processor library.

🤝 Contributing

Contributions are welcome! Please see CONTRIBUTING.md for guidelines on how to contribute to this project.

📜 License

This project is licensed under the Apache License, Version 2.0. See the LICENSE file for details.

Gemini Terms of Services

If you make use of Gemini via the Genai Processors framework, please ensure you review the Terms of Service.