A recent leak on a Chinese tech blog suggests Google is preparing to release Gemini 3.5 Pro, a model that reportedly matches GPT-5.5 in coding benchmarks. More strikingly, the post describes 'Spark', an always-on agent that can autonomously execute transactions, such as making purchases. If accurate, this represents a significant advancement in AI agent capabilities, moving beyond chat to real-world actions. For developers and tech founders, this could reshape how AI assistants are integrated into workflows and e-commerce. The leak, while unverified, aligns with Google's push for more autonomous AI systems. We recommend monitoring official announcements and evaluating the implications for your own AI tooling and agent architectures.
A leaked post claims Google's Gemini 3.5 Pro achieves coding performance on par with GPT-5.5 and introduces 'Spark', a full-time agent capable of making purchases. This signals a major leap in AI agent autonomy and competition in the coding assistant market, relevant for developers and businesses building on AI.