Google Gemini 3.5 Pro Months Behind Schedule as Coding Performance Falls Short
Google's next flagship AI model, Gemini 3.5 Pro, is significantly delayed from its May target as the company struggles to improve coding capabilities to match rivals OpenAI and Anthropic. The company has scrapped its original base model and restarted training, frustrating internal teams concerned about losing competitive ground.
The Delay and Rebuild
Google is months behind schedule on delivering Gemini 3.5 Pro, its most powerful flagship AI model, because the company has been taking time to try to improve its capabilities, particularly in coding. Google DeepMind reportedly scrapped its original model after engineers found structural failures in recursive tool-calling and SVG generation. Pre-training — the initial training run on the vast dataset that gives a model its fundamental capabilities — is the single most expensive phase of building a frontier AI model, making this decision a significant commitment of resources and time.
Competitive Pressure and Market Timing
OpenAI released GPT-5.6 on July 9, 2026, with a new ChatGPT Work product aimed at professional use, while Grok 4.5 opened to the public the same day. Gemini 3.5 Pro release targets July 17, 2026 — four days away — but Google has not confirmed the date, the 2-million-token context window, or pricing. The timing is critical: if you're a CIO deciding what model stack to standardize on for the second half of 2026, July is not an abstract date. Procurement calendars are real. Security reviews are real. Developers start building internal habits around whatever tool is available when the budget clears.
Internal Friction and Talent Exodus
The delay has been a source of frustration for Google engineers, AI researchers and managers, many of whom are concerned the company risks losing an edge in the market as rivals Anthropic and OpenAI produce models that exceed Gemini's capabilities. The delay lands as four senior AI researchers, including Gemini co-lead Noam Shazeer, depart for OpenAI and Anthropic. John Jumper, the Google DeepMind scientist whose AlphaFold work helped win the 2024 Nobel Prize in Chemistry, has also said he's leaving for Anthropic.
Structural Issues Blocking Release
Recursive tool-call stability is the defining requirement for an agentic coding model, which is the use case Google has staked the entire 3.5 generation on. The pressure on Gemini 3.5 Pro sits in coding, tool use, and long-horizon tasks. Google was collecting feedback around agent performance and token consumption, including lessons from Gemini 3.5 Flash. That is exactly where enterprise buyers now test models. They don't care much about a lovely one-paragraph answer if the same model burns through a budget while editing a codebase or loses the thread halfway through a multi-step workflow.