Google’s Flagship AI Model Faces Significant Delays
Google’s highly anticipated AI model, Gemini 3.5 Pro, has reportedly encountered delays of several months due to its inability to satisfy the company’s internal performance benchmarks.
Initially projected for a June launch, the model remains in a state of testing as engineers diligently strive to enhance its capabilities, especially in the domain of software code generation.
This postponement has engendered frustration among engineers, researchers, and management within Google. Many are concerned that the company might be ceding ground to competitors such as OpenAI, Anthropic, and Meta.
In a bid to ameliorate coding performance, Google updated the training data for Gemini 3.5 Pro late last month; however, the outcomes reportedly fell short of anticipated standards.
The ramifications of this setback were felt in the stock market, with Alphabet shares experiencing a decline.
Upon analysis, it appears that Google’s vast corporate framework may be exacerbating these delays, as disparate teams across Google Cloud, DeepMind, Android, and other divisions are developing overlapping AI coding tools.
“We’re shipping quickly across a wide range of models while keeping them highly cost-effective for customers,” stated a spokesperson for Google.
“We’re currently testing 3.5 Pro, an upgraded Flash model, along with other models in collaboration with partners, and we’re actively engaged with the US government on model testing and broader frameworks.”
The complex hierarchy of executives and product teams complicates the release of new Gemini models across various platforms, including Search, YouTube, and Maps.
In response to the challenges, Google has initiated steps to streamline its coding tools. Recently, a dedicated team was established within DeepMind to concentrate on AI-assisted software development.
The company asserts that approximately 75% of its newly generated code is now produced by AI prior to review and approval by engineers. Nevertheless, employees continue to encounter limitations and computing capacity constraints when utilizing internal AI systems.
Gemini 3.5 Flash, introduced at Google I/O in May, has garnered adoption from several companies, including the design platform Figma.
However, some business clients reportedly view the model as slower and more costly than earlier iterations, while still lagging behind premium models provided by competitors.
As it stands, Google has not announced a revised release date for Gemini 3.5 Pro.
Source link: Channelnews.com.au.
