Meta's AI Model Faces Setback: Delayed Release

Meta's next-generation foundational AI model, reportedly codenamed Avocado, has experienced a delay in its release, now anticipated in May according to sources cited by The New York Times. This postponement stems from the model's performance in internal testing, which fell short in areas such as reasoning, coding, and writing when compared to leading models from Google, OpenAI, and Anthropic.

The company has made substantial financial commitments to its AI initiatives, including a $14.3 billion investment for a stake in a startup and appointing its leader as chief AI officer. Meta confirmed plans to increase its total spending from $72 billion in the previous year to a range of $115 billion to $135 billion, primarily to support its AI laboratories.

Despite these significant investments, Meta appears to be trailing behind competitors. Google's Gemini 3 model, released in November, demonstrated impressive coding and research capabilities. OpenAI has also rolled out updates to GPT-5, while Anthropic's Claude Code and Cowork have been recognized for their reliability. Meta's spokesperson expressed excitement for the upcoming release, emphasizing the rapid trajectory of their AI development.

Beyond AI model development, Meta faces other challenges. Renewed privacy concerns surround its Ray-Ban smart glasses, with a lawsuit alleging misuse of recorded footage. Additionally, a high-profile trial is examining the potential addictive nature and health risks posed by Meta's social media platforms, such as Instagram and Facebook, to teenagers.