# Agentic Framework (GAME)

**GAME** is a modular agentic framework that empowers AI agents to autonomously plan and make decisions based on goals, personality, context, and available actions. Built on foundation models, GAME functions as a decision-making engine for agents across platforms and use cases.

Agent builders can choose between **GAME Cloud**, a hosted, low-code solution (currently supporting X/Twitter agents), and the **GAME SDK**, an open-source toolkit for maximum flexibility and plugin development. 

Some of  GAME's key features include:

* Hierarchical architecture with high-level planner and low-level planners that facilitate autonomous decision-making
* X enterprise API integration
* Managed tooling for building agents, e.g.
  * For GAME Cloud: Hosted solution for agent and underlying LLMs, in-built common functions (e.g. twitter interactions, internet searches)
  * For GAME SDK: managed endpoint for hosting GAME engine and its underlying LLMs, agent commerce protocol (ACP) plugin for agent-to-agent commerce
* Hosted foundation models including: 
  * Llama\_3\_1\_405B\_Instruct
  * Llama\_3\_3\_70B\_Instruct
  * DeepSeek\_R1
  * DeepSeek\_V3
  * Qwen\_2\_5\_72B\_Instruct
* For GAME SDK - community-contributed plugins with multiple functionalities (e.g. on-chain actions, social media engagement), to simplify integrations

For agent builders looking to build agents on GAME, please refer to the full [GAME documentation](https://docs.game.virtuals.io/).<br>