Alphabet’s Google unveiled three budget-friendly versions of its Gemini artificial intelligence model on Tuesday, but the company remained silent on when its highly anticipated flagship model will actually arrive.
The newly released lineup consists of Gemini 3.6 Flash and Gemini 3.5 Flash-Lite — updated takes on Google’s lighter-weight AI offerings — designed for everyday tasks that don’t demand the most advanced technology available. Google also introduced a brand-new lightweight option called Gemini 3.5 Flash Cyber, built specifically with cybersecurity uses in mind.
Despite those releases, the tech world’s attention remains fixed on Gemini 3.5 Pro, the premium version of the model that investors and analysts are watching closely as a measure of whether Google’s DeepMind AI lab can hold its own against rivals Anthropic and OpenAI.
At Google’s I/O developer conference back in May, CEO Sundar Pichai announced during his keynote speech that Gemini 3.5 Pro was on track for a June debut. That deadline has come and gone.
The company had made a strong impression last November when it launched the Gemini 3 model family, which many saw as a sign that Google had reclaimed a leading position in the AI race after falling behind for years. That release reportedly prompted OpenAI to declare an internal “code red” to sharpen its focus on its competing GPT model.
Since then, Google’s competitors have been fighting back. Anthropic released its latest Mythos 5 and Fable 5 models — tools so advanced that the U.S. government considered them a potential national security risk. OpenAI, meanwhile, launched GPT-5.6 earlier this month after its own government-requested delay to review security concerns.
Sources say the holdup with Gemini 3.5 Pro stems from the model failing to meet internal benchmarks, especially in coding — an area that has become increasingly valuable for businesses using AI tools.
Pichai has been emphasizing cost savings as a key selling point. He has said businesses could potentially cut their AI spending by more than $1 billion annually by moving the bulk of their AI workloads over to Gemini models.
Google DeepMind staff echoed that message in press materials accompanying Tuesday’s launches, calling the Gemini 3.5 Flash Cyber “a cost-efficient and highly capable alternative to large, costly cybersecurity models.”
As for the flagship model everyone is waiting on, Google said Gemini 3.5 Pro is currently being tested with select partners and will be available “soon.” The company also revealed in one press release that training has already begun on Gemini 4.
Pichai is expected to face pointed questions about Gemini 3.5 Pro’s timeline when Alphabet holds its second-quarter earnings call on Wednesday.
