Google Launches 3 Gemini Models, Starts Gemini 4 Pre-Training
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Google launched Gemini 3.6 Flash, Gemini 3.5 Flash-Lite, and Gemini 3.5 Flash Cyber, positioning them as efficient, low-latency options for building AI agents at scale.
- Google announced it has begun its "most ambitious pre-training run yet" for Gemini 4, which the company describes as a completely new foundation model.
- Gemini 3.6 Flash and Gemini 3.5 Flash-Lite reduce average time per task to approximately 50% of their predecessors, driven by higher token efficiency and faster output speeds (per @artificialanlys).
- Gemini 3.6 Flash cuts cost per task by roughly 18% versus Gemini 3.5 Flash, while Gemini 3.5 Flash-Lite more than doubles in cost compared to 3.1 Flash-Lite (per @artificialanlys).
- Google's Josh Woodward said Gemini 3.6 Flash reduces token usage by up to 65% on complex coding tasks and that Gemini 3.5 Flash-Lite reaches 350 output tokens per second, with both live in the Gemini app.
- Gemini 3.5 Pro remains delayed despite expectations, prompting observers to flag that Google's AI division appears to be falling behind OpenAI and Anthropic (per @shakeelhashim) ahead of Alphabet's earnings report.
Why it matters: Google is shipping cheaper, faster Gemini variants to defend its position in the AI agent market, but the delayed Gemini 3.5 Pro flagship and questions about whether Google has fallen behind OpenAI and Anthropic will face Alphabet's investors directly at its next earnings call. Pre-training Gemini 4 signals a long-term commitment to rebuilding the flagship line rather than patching the current generation.
