Lock.io officially releases FLock Web3 Agent Model

FLock.io officially releases the FLock Web3 Agent Model, a large model optimized for Web3 tasks. It has precise function call capabilities and can efficiently understand and execute complex Web3 tasks.

The model is trained by the FLock community in AI Arena. Users can act as task creators, training nodes, validators, and delegates to jointly promote the evolution of the model. The model is based on Task1, and contributors to the model can receive corresponding rewards.

In web3 type function call benchmark tests, the FLock Web3 Agent Model achieved an accuracy match rate of 75.93%, surpassing GPT-4o, Gemini Flash 2.0 , DeepSeek V3 and Qwen 2.5. The model and related evaluation standards are fully open source to ensure transparency and accessibility.