Meta AI is set to launch a new model called Avocado, with updates already appearing on their website and app. Users can now choose between Fast and Thinking modes and connect apps like Gmail and Google Calendar. The integration of these applications suggests that Meta AI will soon support Multi-Channel Prompts (MCPs), enhancing its functionality. A Memory section has also been added to settings, allowing users to test these new features. The integration of Manus AI hints at new capabilities, including a Manus-style agent and a browser agent, which could significantly enhance user experience. A new Tasks menu is being developed to enable users to schedule recurring prompts, similar to features in other tools. Code traces show that Meta is exploring voice agent support, although these features are still in development. Meta is experimenting with several advanced models, including Gemini, ChatGPT, and Claude, while internal testing reveals multiple development modes for Avocado. Meta's approach appears to focus on closing the feature gap with competitors. A model named Big Brain is also referenced, suggesting parallel processing of multiple model outputs to select the best response. Additionally, Meta is developing a shopping assistant, building on its presence in Facebook and Instagram. An OpenClaw integration is on the horizon, allowing users to utilize their own API keys within Meta AI. This move may open new opportunities for customization and integration, although details on the timing of these releases remain unclear.