ChatArena is an open-source Python framework designed to facilitate the development of multi-agent language game environments for large language models (LLMs). It provides a flexible structure for defining multiple players, environments, and their interactions, enabling researchers and developers to explore and enhance AI communication and collaboration skills. With user-friendly interfaces, including both web UI and command-line options, ChatArena supports the creation of diverse language game scenarios, such as conversational simulations, educational settings, and social deduction games.

chatarena

An open-source framework for creating multi-agent language game environments to enhance AI communication and collaboration capabilities.

ChatArena

Ax is an open-source TypeScript framework designed to facilitate the development of agents and workflows powered by large language models (LLMs). Inspired by Stanford's DSPy paper, Ax provides tools for creating chat interfaces, retrieval-augmented generation (RAG) systems, streaming applications, and multi-modal interactions. It supports various LLMs and vector databases, offers automatic prompt generation from simple signatures, and includes features like output validation during streaming, multi-agent orchestration, and OpenTelemetry tracing for observability. Ax aims to streamline the creation of efficient, production-ready AI applications.

An open-source TypeScript framework for building LLM-powered agents and agentic workflows.

LiteLLM is an open-source toolkit that streamlines interactions with over 100 large language models (LLMs) by providing a unified API in the OpenAI format. It offers both a Python SDK and a proxy server (LLM Gateway) to manage authentication, load balancing, and spend tracking across various LLM providers, including OpenAI, Azure OpenAI, Vertex AI, and Amazon Bedrock. LiteLLM supports features such as retry and fallback logic, rate limiting, and logging integrations with tools like Langfuse and OpenTelemetry. It is designed to simplify the integration of multiple LLMs into applications, ensuring consistent output formats and efficient resource management.

litellm

An open-source Python SDK and proxy server for managing authentication, load balancing, and spend tracking across 100+ LLMs using a unified OpenAI format.

LiteLLM

LiveKit Agents is an open-source framework designed to facilitate the development of AI-driven server programs capable of real-time interaction through voice, video, and data channels. It enables the creation of programmable, multimodal AI agents that can process and generate audio, video, and text, integrating seamlessly with large language models (LLMs) and other AI models. The framework supports flexible integrations, including plugins for popular LLMs, speech-to-text (STT), text-to-speech (TTS), and voice activity detection (VAD) services. LiveKit Agents also offers built-in task scheduling, load balancing, and real-time media transport over WebRTC, making it suitable for applications such as AI voice assistants, call centers, transcription services, and real-time translation.

livekit-agents

An open-source framework for building real-time, multimodal AI applications that can see, hear, and speak.

LiveKit Agents

LlamaGym is an open-source Python framework designed to simplify the fine-tuning of large language model (LLM) agents through online reinforcement learning. By providing a standardized environment similar to OpenAI's Gym, LlamaGym allows developers to efficiently train LLM-based agents by managing conversation context, episode batching, reward assignment, and proximal policy optimization (PPO) setup. This framework enables rapid experimentation with agent prompting and hyperparameters across various Gym environments, facilitating the development of more capable and responsive AI agents.

llamagym

An open-source Python framework for fine-tuning large language model (LLM) agents using online reinforcement learning.

LlamaGym

Marvin is a lightweight AI engineering framework designed to facilitate the development of natural language interfaces that are reliable, scalable, and easy to trust. It provides a suite of tools for common large language model (LLM) use cases, including entity extraction, classification, and synthetic data generation. Marvin's components are simple and self-documenting, allowing developers to integrate AI functionalities into their applications with minimal effort. The framework supports multi-modal inputs, including text, images, and audio, making it versatile for various AI-powered applications.

marvin

An open-source AI engineering framework for building reliable, scalable, and trustworthy natural language interfaces.

Marvin

uAgent is an open-source Python framework developed by Fetch.ai that enables developers to create autonomous AI agents with ease. It provides simple and expressive decorators for scheduling tasks or responding to various events. Upon startup, each agent automatically joins a decentralized network by registering on the Almanac, a smart contract deployed on the Fetch.ai blockchain. This ensures that uAgent messages and wallets are cryptographically secured, protecting their identities and assets.

uagent

An open-source Python framework for creating autonomous AI agents with decentralized capabilities.

uAgent

ControlFlow is an open-source Python framework developed by Prefect for orchestrating AI-powered workflows. It enables developers to break down complex AI tasks into discrete, manageable steps, assign specialized AI agents to each task, and combine these tasks into cohesive flows. ControlFlow integrates seamlessly with large language models (LLMs) like OpenAI's GPT-4 and Anthropic's Claude, allowing for flexible and transparent AI development. Key features include structured results, multi-agent orchestration, and native observability, making it suitable for building robust, intelligent applications.

controlflow

A Python framework for building agentic AI workflows with task-centric design.

ControlFlow

DAGent is an open-source Python framework designed to facilitate the rapid development of AI agents by structuring them into workflows using Directed Acyclic Graphs (DAGs). This approach allows each function to be set up as a node within a graph, enabling complex, multi-step workflows and decision-making processes. By organizing activities and choices in a DAG format, developers can create adaptive and efficient AI agents capable of handling intricate tasks with enhanced reliability and fault tolerance.

dagent

An open-source Python library for creating AI agents structured as Directed Acyclic Graphs (DAGs) to manage decision-making tasks and function executions.

DAGent

TaskWeaver is an open-source, code-first agent framework developed by Microsoft for seamlessly planning and executing data analytics tasks. It interprets user requests through code snippets and efficiently coordinates various plugins to perform data analytics or workflow automation tasks. TaskWeaver supports rich data structures in Python, such as DataFrames, and allows the incorporation of domain-specific knowledge to improve task execution reliability. It also ensures stateful execution, maintaining the execution state across multiple interactions, and provides secure code execution environments.

taskweaver

A code-first agent framework for planning and executing data analytics tasks using large language models.

TaskWeaver

LangGraph is an open-source library designed to facilitate the development of stateful, multi-actor applications utilizing large language models (LLMs). It enables the creation of complex agent and multi-agent workflows, supporting features such as cycles, controllability, and persistence. LangGraph allows developers to define flows involving cycles, essential for most agentic architectures, and provides fine-grained control over both the flow and state of applications. Additionally, it includes built-in persistence, enabling advanced human-in-the-loop and memory features, making it suitable for creating reliable, fault-tolerant agent-based systems.

langgraph

An open-source AI framework for building stateful, multi-actor applications with large language models (LLMs).

LangGraph

LlamaIndex is an open-source data framework designed to facilitate the integration of large language models (LLMs) with custom data sources, enabling the development of AI-powered applications. It provides a flexible structure for building agentic generative AI applications, allowing LLMs to work with data in various formats. LlamaIndex offers state-of-the-art retrieval-augmented generation (RAG) algorithms, reliable integrations across data loading, indexing, and retrieval, and supports the creation of LLM-powered agents capable of performing complex workflows over data and services. With a focus on production readiness, LlamaIndex serves as a comprehensive solution for developers aiming to build knowledge assistants and other AI applications over enterprise data.

llamaindex

An open-source data framework that enables large language models to interact with custom data sources for building AI applications.

AI Agents for: AI researchers