At a conference in Bangkok, NEAR Protocol announced plans to create the largest open-source AI model. The new 1.4 trillion-parameter model will surpass Meta's current open-source Llama model.
NEAR Protocol's New AI Model
The development will be driven by crowdsourced research and contributions from thousands of participants globally. Access to development will be provided through the NEAR AI Research Hub, starting with a 500 million-parameter model. This aims to include a wide range of participants and allows for cooperative project advancement.
Monetization and Privacy Protection
One important aspect is the project's business model. NEAR Protocol plans to monetize the model through token sales, aiming to secure $160 million to fund training and computation. To safeguard participant privacy, Encrypted Trusted Execution Environments (TEEs) will be used. A reward system is also in place to sustain contributions and technological advancements.
Benefits of AI Assistant
An AI Assistant was also introduced at the conference, showcasing the capabilities of NEAR's AI infrastructure. The assistant performs various tasks like creating memecoins, designing websites, and finding service providers. Future versions will expand functionality, including booking flights and interacting with Web2 services via blockchain. A key feature is its chain abstraction technology, enabling asset swaps across blockchains like Bitcoin, Ethereum, and NEAR Protocol.
NEAR Protocol's plans for creating the largest AI model are ambitious, involving global cooperation. This reflects their vision of merging traditional web services with decentralized technology.
