According to Beating's monitoring, Google's Gemini 3.5 Pro, originally scheduled for June release, has been delayed for several months as its coding capabilities fell short of internal benchmarks. After updating training data in late June, the model's performance remained unsatisfactory.
Internal sources cited by APPSO indicated that the latest evaluation results may be inferior to Meta's Muse Spark. Some associated team members lost test access, with only the model team retaining viewing rights to evaluation metrics. Google confirmed the model is still under testing with partners and testing an upgraded Flash version.