What happened
Google DeepMind released Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber on Tuesday. Gemini 3.6 Flash reduces token usage by up to 17% compared to its predecessor 3.5 Flash while improving coding and multimodal capabilities. 3.5 Flash Cyber, a cybersecurity-focused model, will be exclusively available to governments and trusted partners through a limited access pilot program.
Why it matters
The launch emphasizes efficiency and speed for AI agents running at scale, addressing a core need for production systems. However, Google notably did not release the long-anticipated Gemini 3.5 Pro update—last updated in February—signaling internal delays as competitors including OpenAI (which has released GPT-5.5 and begun rolling out GPT-5.6) and Anthropic (which launched Claude Opus 4.8, Claude Sonnet 5, and expanded Fable 5 access) maintain a faster release cadence.
What to watch
Google DeepMind product lead Logan Kilpatrick said Tuesday the company is currently testing Gemini 3.5 Pro with partners and hopes to "land soon." The team has also started its most ambitious pre-training run yet for Gemini 4, according to Kilpatrick.
Summaries like this, in your inbox every morning.
Google's Tuesday release reflects a strategic pivot toward efficiency and operational scale at a moment when its competitors are accelerating their own cadence. While the company shipped three models—each addressing a specific use case (cost-effectiveness, faster response, and cybersecurity)—the absence of Gemini 3.5 Pro reveals internal constraints. The article cites Bloomberg's reporting of performance-goal misses, suggesting Google's engineering timeline has slipped. In the interim, OpenAI has moved beyond GPT-5.5 to begin rolling out GPT-5.6, and Anthropic has launched multiple frontier models including Claude Opus 4.8 and expanded access to Fable 5. This competitive backdrop frames Google's focus on cheaper, faster inference as a near-term strategy: serving production customers at scale while flagship-model development continues internally. Logan Kilpatrick's statement that the team has begun "its most ambitious pre-training run yet for Gemini 4" signals that Google's long-term roadmap remains ambitious, even as near-term releases lag rival labs' pace.
For example, today's edition would include:
AI-summarized, only the topics you pick — one digest a day via Email, LINE, or Slack.
Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. Q&As are published on this page for other readers too.
DeepSeek CEO Liang Wenfeng told investors the startup aims to rely on Huawei chips to train its models
Microsoft opened a cloud region in Hyderabad, turning part of its US$20.5 billion India commitment into operat…
Arm estimates agentic AI could raise data-center CPU demand from roughly 30 million cores to about 120 million…
Shopify is partnering with Meta so Muse, Meta's AI shopping agent, can complete purchases at Shopify merchants…
Simon Willison released llm-typesafe 0.1a0, an LLM plugin adding support for TypeSafe AI's Jev model, installe…
NVIDIA released Isaac ROS 5.0, adding agentic workflows, Isaac Skills for setup and manipulation, and support…
