Overview
- Bloomberg reported Thursday that Gemini 3.5 Pro, which Google had said would arrive in June, is now months late after internal tests fell short of the company’s performance goals.
- Google told Reuters it is testing 3.5 Pro and other upgraded models with partners and is productively engaged with the U.S. government about the systems.
- A late-June update to the training data did not deliver expected gains, with sources saying the model’s coding ability was a particular shortfall compared with internal targets.
- Rivals have continued to push new models into the market, including OpenAI’s recent GPT-5.6 and Anthropic’s restored Mythos/Fable offerings, raising competitive pressure on Google’s flagship release.
- Alphabet shares fell after the reports, and the delay could slow enterprise rollouts, change customer buying decisions for coding tools, and prolong regulatory checks on powerful AI systems.