Google’s next flagship AI model missed its own launch window, and reported coding problems appear to be the reason.
Gemini 3.5 Pro was supposed to arrive in June. Instead, Google is still testing the model while trying to improve its performance, particularly its ability to write code, according to reporting from Reuters based on a Bloomberg investigation.
June came and went. Gemini 3.5 Pro did not.
Google has not announced a new release date. A spokesperson says the company is currently testing Gemini 3.5 Pro, an upgraded Flash model, and other systems with partners.
Google Promised Gemini 3.5 Pro for June
Google introduced the Gemini 3.5 family at Google I/O on May 19, 2026. Gemini 3.5 Flash launched immediately, while CEO Sundar Pichai said Pro was already being used internally and would arrive the following month.
“We are using it internally, it’s showing great improvements, and it will be coming next month,” Pichai said in Google’s official I/O keynote transcript.
Google repeated that timeline in its I/O 2026 announcement roundup.
That deadline has now been missed by several weeks. Bloomberg’s reporting goes further, describing the model as months behind its internal schedule rather than merely a few weeks late to the public.
Coding Performance Is Reportedly Holding It Back
The main issue appears to be coding.
Google reportedly updated the data used to train Gemini in late June to improve its programming abilities. The results fell short of internal expectations, according to Bloomberg’s reporting, which was also summarized by 9to5Google.
That matters because coding has become one of the clearest battlegrounds in the frontier-model race. Developers increasingly judge models by how well they can navigate repositories, fix bugs, use tools, complete long-running tasks, and operate inside agentic coding environments.
A model can look strong on general reasoning benchmarks and still lose developers if it struggles with practical software work.
Google has already positioned Gemini 3.5 as a family built for agents and complex workflows. The company says Flash delivers strong coding and agentic performance while remaining faster and cheaper than larger frontier models. It has also pushed Flash deeper into agent workflows with built-in computer use.
Pro is supposed to be the more capable flagship version of that generation. Releasing it with weaker-than-expected coding performance would undercut the pitch before developers even finished their first benchmark run.
The Delay Is Creating Pressure Inside Google
The reported delay has frustrated some Google engineers, researchers, and managers. Bloomberg’s sources said there are concerns that competing labs are moving faster while Gemini 3.5 Pro remains in development.
That does not mean Google has fallen out of the AI race. Gemini is integrated across Search, Android, Workspace, Google Cloud, and the company’s developer platforms. Google also controls its own models, infrastructure, consumer products, and distribution at a scale few competitors can match.
But model momentum matters.
Frontier AI development now moves in compressed cycles. A model delayed by a few months can launch into a completely different competitive landscape than the one it was built to enter. Moonshot AI’s Kimi K3 launch is one recent example of how quickly another serious model can appear and seize attention.
Every additional delay gives rival systems more time to improve, attract developers, and become embedded in real workflows.
Google Says It Is Still Shipping Quickly
Google has not confirmed the specific internal performance problems described in Bloomberg’s report.
A company spokesperson told Reuters that Google is testing Gemini 3.5 Pro, an upgraded Flash model, and other models with partners. The spokesperson also said Google is shipping quickly across a wide range of models while keeping them cost-effective for customers.
That response confirms the model remains in active testing. It does not answer the two questions developers care about most: when Gemini 3.5 Pro will launch and how much better it will be than the models already available.
Holding the release may ultimately be the correct decision. Shipping a flagship model before it is ready could cause more damage than missing a deadline, especially when developers can immediately compare its coding performance against established alternatives.
Still, Google set the June expectation itself.
What Happens Next
The next signal will be whether Google provides a revised launch window or continues testing Gemini 3.5 Pro until it reaches the company’s internal targets.
Google could narrow the gap through an upgraded Gemini 3.5 Flash release before Pro arrives. Flash is no longer merely the lightweight member of the family. Google is positioning it as a serious model for coding, computer use, and agentic workloads.
But the flagship still matters. Gemini 3.5 Pro is supposed to represent Google’s best work and compete directly at the top of the market.
For now, the facts are simple: Google promised Gemini 3.5 Pro in June, the model missed that deadline, and reported attempts to improve its coding performance have not yet produced a release-ready result.
Google can afford to miss a date. It cannot afford to ship a flagship that loses the coding comparison on day one.
Sources
- Reuters: Google Gemini launch delayed as technology falls short of internal goals
- Google: Sundar Pichai’s I/O 2026 keynote
- Google: 100 things announced at Google I/O 2026
- 9to5Google: Gemini 3.5 Pro delayed over coding performance
