Meta Delays Next-Gen AI Model "Avocado" to May: Performance Falls Short of Rivals, Considers Licensing Google Gemini — $135B AI Bet Under Scrutiny
Meta Platforms has delayed the release of its next-generation AI model codenamed "Avocado" from March to at least May 2026, after internal testing revealed the model's performance falls short of leading competitors including OpenAI's GPT, Anthropic's Claude, and Google's latest Gemini models. The New York Times first reported that Meta's AI division leaders are now considering temporarily licensing Google's Gemini to power Meta's AI products. The delay comes as Meta plans to invest up to $135 billion in AI infrastructure in 2026 and reportedly considers layoffs of 20% or more of its workforce to fund aggressive AI spending.

Zhipu Launches GLM-5-Turbo: World's First OpenClaw-Optimized Foundation Model, 20% API Price Increase, "Lobster Plan" Starting at ¥39 — The AI Agent Model Wars Escalate
Zhipu AI (02513.HK) officially released GLM-5-Turbo on March 16, 2026 — the world's first foundation model specifically optimized for OpenClaw ("Lobster") scenarios. The model is enhanced from the training stage for core Agent capabilities including tool calling, instruction following, scheduled tasks, and long-chain execution. Simultaneously, Zhipu raised GLM-5-Turbo API prices by 20% and launched "Lobster Plan" subscriptions starting at ¥39 for 35 million tokens. GLM-5-Turbo has been integrated into iSoftStone's Mechrevo "Lobster Box," creating a native AI Agent terminal experience. Zhipu shares surged up to 16% intraday following the announcement.





