[Gemini 3.5 Pro Delayed for Months, Rumored to Perform Worse Than Muse Spark] According to monitoring by Beating, Google's flagship model Gemini 3.5 Pro has been delayed for several months. Originally scheduled for release in June, the model's coding capabilities have yet to meet internal targets. Even after updating training data in late June, the results remain unsatisfactory. APPSO cited internal sources claiming that the latest evaluations of 3.5 Pro might even fall short of Meta's Muse Spark. Recently, testing permissions for some employees in related business units have been revoked, leaving only the model team with access to continue using and reviewing evaluation results. APPSO also reported that morale within the Gemini-related teams is low. Some employees, worried about team restructuring and layoffs, have started considering joining other AI labs. Recently, Noam Shazeer moved to OpenAI, and AlphaFold's core researcher John Jumper also left Google to join Anthropic. However, there is currently no reliable evidence to support claims that 3.5 Pro has been canceled. Google responded by stating that the model is still being tested with partners, while upgraded models like Flash are also undergoing testing. [Original Link]