Author: Deep Value Memetics, Translation: Golden Finance Xiaozou

In this article, we will explore the prospects of the Crypto X AI frameworks. We will focus on the current four major frameworks (ELIZA, GAME, ARC, ZEREPY) and their technical differences.

1. Introduction

In the past week, we have researched and tested the four major Crypto X AI frameworks: ELIZA, GAME, ARC, and ZEREPY. Our conclusions are as follows.

We believe AI16Z will continue to dominate. The value of Eliza (with a market share of about 60% and a market value exceeding $1 billion) lies in its first-mover advantage (Lindy effect) and its growing adoption by developers, evidenced by data such as 193 contributors, 1,800 forks, and over 6,000 stars, making it one of the most popular code repositories on GitHub.

So far, the development of GAME (with a market share of about 20% and a market value of about $300 million) has been very smooth, rapidly gaining adoption. As VIRTUAL just announced, the platform has over 200 projects, 150,000 daily requests, and a weekly growth rate of 200%. GAME will continue to benefit from the rise of VIRTUAL and will become one of the biggest winners in its ecosystem.

Rig (ARC, with a market share of about 15% and a market value of about $160 million) is particularly notable for its modular design that is very easy to operate and can dominate as a pure play in the Solana ecosystem (RUST).

Zerepy (with a market share of about 5% and a market value of about $300 million) is a relatively niche application specifically targeting the enthusiastic ZEREBRO community, and its recent collaboration with the ai16z community may yield synergies.

We note that our market share calculations cover market value, development records, and underlying operating system terminal markets.

We believe that in this market cycle, the framework segment will be the fastest-growing area, and the total market value of $1.7 billion could easily grow to $20 billion, which remains relatively conservative compared to the peak valuation of L1 in 2021, when many L1 valuations exceeded $20 billion. Although these frameworks serve different end markets (chains/ecosystems), given that we believe this field is on a continuous upward trend, a market cap-weighted approach may be the most prudent.

2. Four Major Frameworks

In the table below, we list the key technologies, components, and advantages of each major framework.

(1) Framework Overview

In the cross-field of AI X Crypto, several frameworks facilitate AI development. They are AI16Z's ELIZA, ARC's RIG, ZEREPY's ZEREBRO, and GAME's VIRTUAL. Each framework meets different needs and philosophies in the AI agent development process, ranging from open-source community projects to performance-focused enterprise-level solutions.

This article first introduces the frameworks, explaining what they are, the programming languages, technology architectures, algorithms they use, their unique features, and potential use cases for the frameworks. We then compare each framework in terms of usability, scalability, adaptability, and performance, exploring their respective strengths and limitations.

ELIZA (developed by ai16z)

Eliza is a multi-agent simulation open-source framework designed to create, deploy, and manage autonomous AI agents. It is developed in the TypeScript programming language, providing a flexible and scalable platform for building intelligent agents that can interact with humans across multiple platforms while maintaining consistent personalities and knowledge.

The core features of the framework include a multi-agent architecture that supports the simultaneous deployment and management of multiple unique AI personalities, a role system for creating different agents using a character file framework, and memory management capabilities that provide long-term memory and context awareness through advanced retrieval-augmented generation (RAG) systems. Additionally, the Eliza framework offers smooth platform integration for reliable connections with Discord, X, and other social media platforms.

From the perspective of AI agent communication and media capabilities, Eliza is an excellent choice. In communication, the framework supports integration with Discord's voice channel functionality, X functionality, Telegram, and API direct access for customized use cases. On the other hand, the framework's media processing capabilities extend to PDF document reading and analysis, link content extraction and summarization, audio transcription, video content processing, image analysis, and dialogue summarization, effectively handling various media inputs and outputs.

The Eliza framework provides flexible AI model support through local inference of open-source models, cloud inference from OpenAI, and default configurations (such as Nous Hermes Llama 3.1B), and integrates support for Claude to handle complex tasks. Eliza adopts a modular architecture with broad operating system, custom client support, and a comprehensive API, ensuring scalability and adaptability between applications.

Eliza's use cases span multiple domains, such as AI assistants for customer support, community moderation, and personal tasks, as well as content auto-creators, interactive bots, and brand representatives in social media roles. It can also act as a knowledge worker, taking on roles such as research assistants, content analysts, and document processors, and supports interactive roles like role-playing bots, educational mentors, and agent representatives.

Eliza's architecture is built around the agent runtime, which seamlessly integrates with its role system (supported by model providers), memory manager (connected to a database), and operating system (linked to platform clients). Unique features of the framework include a plugin system that supports modular functional extensions, enabling multimodal interactions such as voice, text, and media, and compatibility with leading AI models (e.g., Llama, GPT-4, and Claude). With its diverse capabilities and robust design, Eliza stands out as a powerful tool for developing AI applications across various domains.

G.A.M.E (developed by Virtuals Protocol)

The Generative Autonomous Multimodal Entity framework (G.A.M.E) aims to provide developers with API and SDK access for AI agent experimentation. This framework offers a structured approach to managing the behavior, decision-making, and learning processes of AI agents.

Its core components are as follows: First, the Agent Prompting Interface is the entry point for developers to integrate GAME into agents and access agent behavior. The Perception Subsystem initiates a session by specifying parameters such as session ID, agent ID, user, and other relevant details.

It synthesizes incoming information into a format suitable for the Strategic Planning Engine, serving as the sensory input mechanism for the AI agent, whether in the form of conversation or reactions. At its core is a dialogue processing module that handles messages and responses from agents and collaborates with the perception subsystem to effectively interpret and respond to input.

The Strategic Planning Engine works alongside the dialogue processing module and on-chain wallet operators to generate responses and plans. The engine's functions have two levels: as a high-level planner, it creates broad strategies based on context or goals; as a low-level strategist, it translates these strategies into actionable plans, which are further divided into action planners for specific tasks and plan executors for task execution.

Another independent but important component is the World Context, which refers to the environment, global information, and game state, providing necessary context for the agent's decision-making. Additionally, the Agent Repository is used to store long-term attributes such as goals, reflections, experiences, and personality, which together shape the agent's behavior and decision-making process.

The framework employs short-term working memory and long-term memory processors. Short-term memory retains relevant information about past behaviors, results, and current plans. In contrast, the long-term memory processor extracts key information based on criteria such as importance, recency, and relevance. Long-term memory stores knowledge about the agent's experiences, reflections, dynamic personality, world context, and working memory to enhance decision-making and provide a foundation for learning.

The learning module generates general knowledge from data provided by the perception subsystem, which is fed back into the system to improve future interactions. Developers can input feedback about actions, game states, and sensory data through the interface to enhance the learning capabilities of the AI agent, improving its planning and decision-making abilities.

The workflow begins with developers interacting through the Agent Prompting Interface. Input is processed by the perception subsystem and forwarded to the dialogue processing module, which manages the interaction logic. The Strategic Planning Engine then formulates and executes plans based on this information, utilizing high-level strategies and detailed action plans.

Data from the world context and agent repository informs these processes, while working memory tracks immediate tasks. Meanwhile, the long-term memory processor stores and retrieves long-term knowledge. The learning module analyzes results and integrates new knowledge into the system, enabling continuous improvement of the agent's behavior and interactions.

RIG (developed by ARC)

Rig is an open-source Rust framework designed to simplify the development of large language model applications. It provides a unified interface for interacting with multiple LLM providers (such as OpenAI and Anthropic) and supports various vector storage options, including MongoDB and Neo4j. The unique aspect of the framework's modular architecture lies in its core components, such as the Provider Abstraction Layer, Vector Store Integration, and Agent System, to facilitate seamless interaction with LLMs.

Rig's primary audience includes developers building AI/ML applications using Rust, followed by organizations seeking to integrate multiple LLM providers and vector storage into their Rust applications. The repository utilizes a workspace architecture with multiple crates, supporting scalability and efficient project management. Key features include a provider abstraction layer that standardizes completion and embedding APIs across different LLM providers with consistent error handling. The Vector Store Integration component provides an abstract interface for multiple backends and supports vector similarity search. The agent system simplifies LLM interaction and supports retrieval-augmented generation (RAG) and tool integration. Additionally, the embedding framework also provides batch processing capabilities and type-safe embedding operations.

Rig employs multiple technological advantages to ensure reliability and performance. Asynchronous operations leverage Rust's asynchronous runtime to efficiently handle large numbers of concurrent requests. The framework's inherent error handling mechanism enhances the recovery capability from failures in AI providers or database operations. Type safety can prevent errors during the compilation process, thus enhancing code maintainability. Efficient serialization and deserialization processes support data processing in formats such as JSON, which is crucial for AI service communication and storage. Detailed logging and detection further aid in debugging and monitoring applications.

The workflow of Rig begins when a client initiates a request, which interacts with the appropriate LLM model through the provider abstraction layer. Data is then processed by the core layer, where agents can use tools or access context through vector storage. Responses are generated and refined through complex workflows (such as RAG) before being returned to the client, a process involving document retrieval and contextual understanding. The system integrates multiple LLM providers and vector storage, adapting to model availability or performance updates.

Rig's use cases are diverse, including question-and-answer systems that retrieve relevant documents to provide accurate responses, document search and retrieval systems for efficient content discovery, and chatbots or virtual assistants providing context-aware interactions for customer service or education. It also supports content generation, enabling the creation of texts and other materials based on learning patterns, making it a versatile tool for developers and organizations.

Zerepy (developed by ZEREPY and blorm)

ZerePy is an open-source framework written in Python, designed to deploy agents on X using OpenAI or Anthropic LLM. A modular version derived from the Zerebro backend, ZerePy allows developers to launch agents with core functionalities similar to Zerebro. While the framework provides the foundation for agent deployment, fine-tuning models is essential for generating creative output. ZerePy simplifies the development and deployment of personalized AI agents, particularly targeting content creation on social platforms, nurturing an AI-driven creative ecosystem focused on art and decentralized applications.

The framework is developed using Python, emphasizing agent autonomy and focusing on creative output generation, consistent with ELIZA's architecture and its collaborative relationship with ELIZA. Its modular design supports memory system integration, facilitating agent deployment on social platforms. Key features include a command-line interface for agent management, integration with Twitter, support for OpenAI and Anthropic LLMs, and a modular connection system for enhanced functionality.

ZerePy's use cases cover social media automation, where users can deploy AI agents for posting, replying, liking, and retweeting, thereby enhancing platform engagement. Additionally, it caters to content creation in fields such as music, memes, and NFTs, making it an important tool for digital art and blockchain-based content platforms.

(2) Comparison of the Four Major Frameworks

In our view, each framework offers a unique approach to AI development that aligns with specific needs and environments, shifting our focus from the competitive relationships of these frameworks to their uniqueness.

ELIZA stands out with its user-friendly interface, especially for developers familiar with JavaScript and Node.js environments. Its comprehensive documentation helps set up AI agents across various platforms, although its extensive feature set may present a learning curve. Developed using TypeScript, Eliza is an ideal choice for building agents embedded in the web, as most web infrastructure front-ends are developed using TypeScript. The framework is known for its multi-agent architecture, allowing different AI personalities to be deployed on platforms such as Discord, X, and Telegram. Its advanced memory management RAG system makes it particularly effective for AI assistants in customer support or social media applications. While it offers flexibility, strong community support, and consistent cross-platform performance, it is still in the early stages and may pose a learning curve for developers.

GAME is designed specifically for game developers, providing a low-code or no-code interface through APIs, allowing users with less technical background in the gaming field to utilize it. However, its focus on game development and blockchain integration may present a steep learning curve for those without relevant experience. It excels in procedural content generation and NPC behavior but is limited by the increased complexity associated with its niche and blockchain integration.

Due to the use of Rust, Rig may be less user-friendly given the complexity of the language, presenting significant learning challenges, but for those proficient in system programming, it offers intuitive interaction. Compared to TypeScript, this programming language is known for performance and memory safety. It features strict compile-time checks and zero-cost abstractions, which are essential for running complex AI algorithms. The language is highly efficient, and its low-level control makes it an ideal choice for resource-intensive AI applications. The framework provides high-performance solutions with a modular and scalable design, making it suitable for enterprise applications. However, for developers unfamiliar with Rust, using Rust inevitably faces a steep learning curve.

ZerePy leverages Python, providing high availability for creative AI tasks, with a lower learning curve for Python developers, especially those with an AI/ML background, and benefits from strong community support due to the Zerebro crypto community. ZerePy excels in creative AI applications like NFTs, positioning itself as a powerful tool for digital media and art. While it thrives in creativity, its scope is relatively narrow compared to other frameworks.

In terms of scalability, ELIZA has made significant progress in its V2 update, introducing a unified messaging line and scalable core framework that supports effective management across multiple platforms. However, without optimization, managing this multi-platform interaction may pose scalability challenges.

GAME excels in the real-time processing required for gaming, and its scalability is managed through efficient algorithms and potential blockchain distributed systems, although it may be limited by specific game engines or blockchain networks.

The Rig framework leverages Rust's scalability performance, designed for high-throughput applications, which is particularly effective for enterprise-level deployments, although achieving true scalability may require complex setups.

Zerepy's scalability is aimed at creative output and is supported by community contributions, but its focused core may limit its application in a broader AI environment. Scalability may be challenged by the diversity of creative tasks rather than the number of users.

In terms of adaptability, ELIZA leads with its plugin system and cross-platform compatibility, while its GAME in gaming environments and Rig in handling complex AI tasks are also impressive. ZerePy shows high adaptability in creative fields but is less suitable for broader AI applications.

In terms of performance, ELIZA is optimized for fast social media interactions, with rapid response times being key, but its performance may vary when handling more complex computational tasks.

GAME, developed by Virtual Protocol, focuses on high-performance real-time interactions in gaming scenarios, leveraging efficient decision-making processes and potential blockchain for decentralized AI operations.

The Rig framework, based on the Rust language, provides excellent performance for high-performance computing tasks, suitable for enterprise applications where computational efficiency is critical.

Zerepy's performance is tailored for the creation of creative content, with its metrics centered around the efficiency and quality of content generation, which may be less universal outside the creative field.

The strength of ELIZA lies in its flexibility and scalability, with a high degree of adaptability through its plugin system and role configurations, advantageous for cross-platform social AI interactions.

GAME offers unique real-time interaction capabilities in games, enhanced by blockchain integration that fosters novel AI engagement.

Rig's advantages lie in its performance and scalability for enterprise AI tasks, focusing on providing clean modular code for the health of long-term projects.

Zerepy excels at fostering creativity, leading in digital art AI applications, and is supported by a vibrant community-driven development model.

Each framework has its limitations. ELIZA is still in its early stages, with potential stability issues and a learning curve for new developers. The niche focus of Game may limit broader applications, and blockchain adds complexity. Rig may deter some developers due to its steep learning curve from Rust, while Zerepy's narrow focus on creative output may limit its use in other AI fields.

(3) Framework Comparison Summary

Rig (ARC):

Language: Rust, focusing on safety and performance.

Use Case: Ideal for enterprise-level AI applications due to its emphasis on efficiency and scalability.

Community: Less community-driven, more focused on technical developers.

Eliza (AI16Z):

Language: TypeScript, emphasizing the flexibility of web3 and community involvement.

Use Case: Designed for social interactions, DAOs, and trading, with a particular emphasis on multi-agent systems.

Community: Highly community-driven, with extensive GitHub participation.

ZerePy (ZEREBRO):

Language: Python, making it accessible to a broader base of AI developers.

Use Case: Suitable for social media automation and simpler AI agent tasks.

Community: Relatively new, but expected to grow due to the popularity of Python and support from AI16Z contributors.

GAME (VIRTUAL):

Focus: Autonomous, adaptive AI agents that can evolve based on interactions in virtual environments.

Use Case: Most suitable for AI agent learning and adaptation scenarios, such as games or virtual worlds.

Community: An innovative community, but still determining its positioning in the competition.

3. GitHub Star Data Trends

The above figure shows the GitHub star follow data since the release of these frameworks. Notably, GitHub stars are indicators of community interest, project popularity, and perceived value of the project.

ELIZA (Red Line):

Starting from a low base in July and then significantly increasing the number of stars by late November (reaching 61,000 stars), this indicates a rapid increase in interest, attracting the attention of developers. This exponential growth suggests that ELIZA has gained tremendous traction due to its features, updates, and community engagement. Its popularity far exceeds other competitors, indicating strong community support and broader applicability or interest in the AI community.

RIG (Blue Line):

Rig is the oldest among the four frameworks, with a moderate but consistently increasing number of stars. It has already reached 1,700 stars, but is likely to increase significantly in the coming month. Ongoing development, updates, and a steadily growing user base are reasons for the accumulation of user interest. This may reflect that the framework has a niche user base or is still accumulating reputation.

ZEREPY (Yellow Line):

ZerePy was just launched a few days ago and has already accumulated 181 stars. It is worth emphasizing that ZerePy needs more development to enhance its visibility and adoption rates. Cooperation with AI16Z may attract more code contributors.

GAME (Green Line):

This project has the fewest stars. Notably, this framework can be directly applied to agents in the virtual ecosystem via API, eliminating the need for visibility on GitHub. However, this framework was only made publicly available to builders a little over a month ago, and over 200 projects are currently using GAME to build.

4. Framework Bullish Arguments

Eliza's V2 version will integrate the Coinbase agent suite. All projects using Eliza will support native TEE in the future, allowing agents to operate in secure environments. One upcoming feature of Eliza is the Plugin Registry, enabling developers to seamlessly register and integrate plugins.

Additionally, Eliza V2 will support automated anonymous cross-platform messaging. The tokenomics whitepaper is scheduled to be released on January 1, 2025, which is expected to positively impact the underlying AI16Z token of the Eliza framework. AI16Z plans to continue enhancing the framework's utility and attracting high-quality talent, and the efforts of its main contributors have already proven its capability.

The GAME framework provides no-code integration for agents, allowing the simultaneous use of GAME and ELIZA within a single project, each serving a specific purpose. This approach is expected to attract builders who focus on business logic rather than technical complexity. Although the framework has only been publicly released for about 30 days, it has made substantial progress with the team's efforts to attract more contributors. All projects expected to launch on VIRTUAL will utilize GAME.

Represented by the ARC token, Rig has tremendous potential, although its framework is still in the early growth stage, and the plans to drive project adoption have only been online for a few days. However, high-quality projects adopting ARC are expected to emerge soon, similar to the Virtual flywheel, but with a focus on Solana. The team is optimistic about the partnership with Solana, likening the relationship between ARC and Solana to that of Virtual to Base. Notably, the team not only encourages new projects to use Rig for launch but also encourages developers to enhance the Rig framework itself.

Zerepy is a newly launched framework that is gaining increasing attention due to its collaboration with Eliza. The framework attracts contributors from Eliza, who are actively improving it. Driven by ZEREBRO fans, it has a passionate following and offers new opportunities for Python developers who previously lacked representation in the competition for AI infrastructure. The framework is poised to play an important role in AI creativity.