A new open-source AI model, Molmo, has been released by the Allen Institute for AI (Ai2), enabling developers to create AI agents that can perform complex tasks on computers. The multimodal model can interpret images and converse through a chat interface, potentially helping AI agents browse the web, navigate file directories, and draft documents. With its release, Ai2 aims to empower next-generation apps and make AI more accessible to researchers and startups. The model’s openness and flexibility allow developers to fine-tune it for specific tasks, such as working with spreadsheets, providing a significant advantage over current models like GPT-4, which are limited in their customization options.
Source: https://www.wired.com/story/molmo-open-source-multimodal-ai-model-allen-institute-agents/