TechTimes reports that Google is targeting July 17 for Gemini 3.5 Pro after a full architectural rebuild. Engineers reportedly scrapped the original base model when they found structural failures in recursive tool-calling and SVG generation, delaying the launch to redo the work. Reported specs point to a 2-million-token context window and a Deep Think reasoning layer for multi-step logic, though TechTimes stresses Google has not confirmed the date, the context size, or pricing, and no official model card exists yet. If it lands as described, the model would go head-to-head with OpenAI's GPT-5.6 and Anthropic's Fable 5. Treat everything but the target date as leaks until Google publishes details.