Bloomberg reported on July 16 that Google is delaying the release of its next-generation flagship AI model, Gemini 3.5 Pro, by several months — and that the core reason is coding performance. At its I/O 2026 developer conference in May, Google unveiled Gemini 3.5 Flash and said the Pro version would arrive in June. That deadline passed with no update.
What Went Wrong — Coding
The reason for the delay is clear. Per Bloomberg, Google is "taking time to try to improve [Gemini 3.5 Pro's] capabilities, particularly in coding." In late June, Google updated the data used to train the model in an attempt to sharpen its coding skills, but the results were reportedly "disappointing." That is a sharp contrast to I/O, where Google said the model was "showing great improvements" — suggesting development was effectively reset between the announcement and the missed launch.
Pro originally promised June 2026 (delayed)
Current flagship Gemini 3.1 Pro (February 2026)
Share of new internal code that is AI-generated 75% (April 2026, up from 50% last fall)
The Delay, on a Timeline
| When | What |
|---|---|
| February 2026 | Gemini 3.1 Pro released (current flagship) |
| May 2026 | 3.5 Flash unveiled at I/O 2026; Pro promised for "June" |
| Late June 2026 | Training data adjusted to improve coding → disappointing results |
| July 2026 | 3.5 Pro and an upgraded Flash in testing with partners; no public date |
In a statement, Google said it is "currently testing 3.5 Pro, an upgraded Flash model, and other models with partners," adding that it is "shipping quickly across a wide range of models while keeping them highly cost-effective for customers." In other words, the launch isn't scrapped — the model is being polished before release.
Why Coding Is the Battleground
Coding has become a central front in the frontier-AI race, and Google's own reliance on AI coding tools is rising fast. CEO Sundar Pichai said at Cloud Next in April that 75% of new code at Google is now AI-generated and approved by engineers — up sharply from 50% last fall. That makes coding performance both a product differentiator and a direct lever on the company's own development productivity.
The Internal Picture
Bloomberg's report also captures unease inside Google. Some engineers worry the company is falling behind faster rivals, while others hold a purist stance — that important code should be human-written to meet Google's standards. On top of that, internal AI tools are running into compute-capacity constraints, and an effort is underway to unify the company's scattered internal AI coding tools. DeepMind (AI Studio), Cloud (Vertex), and the Android team (Android Studio) have each run their own efforts, leaving structural fragmentation to resolve.
What It Means
With Gemini 3.1 Pro dating to February, the 3.5 Pro slip meaningfully widens the gap in Google's flagship lineup. The question is how Google balances polish against speed. Choosing to secure coding reliability before shipping — rather than rushing out an unfinished model — is reasonable, but if rivals lock in the developer ecosystem in the meantime, the gap could become hard to reverse.
· Bloomberg — Google Gemini launch delayed as tech falls short of internal goals (7/16, original report)
· 9to5Google — Gemini 3.5 Pro delays due to coding performance (7/16)
· Search Engine Journal — Google Delays Gemini 3.5 Pro Over Coding Issues
· Neowin — Gemini 3.5 Pro faces delays over coding performance misses
- Google delayed Gemini 3.5 Pro, the flagship it promised for June at I/O
- The core reason is coding performance missing internal targets (Bloomberg)
- A late-June training-data adjustment produced disappointing results
- Current flagship 3.1 Pro dates to February — a prolonged top-tier gap
- Developer-market momentum is at stake as Anthropic and OpenAI push coding models