Google Releases Gemini 3.6 Flash; Flagship 3.5 Pro Still Delayed
The short answer
The fast-and-cheap tier keeps improving while Google's answer to GPT-5.6 stays in testing — a pattern worth understanding. Notice the pattern: across every major lab, the fast/cheap tier is improving faster than the flagship tier is shipping.
- Gemini
- model releases
- cybersecurity
What happened
Google announced Gemini 3.6 Flash — faster and cheaper than its predecessor — alongside Gemini 3.5 Flash Cyber, its first model tuned specifically for cybersecurity work. Citing dual-use risk, the Cyber model launches only as a limited pilot inside DeepMind's CodeMender agent for trusted partners and governments.
Meanwhile, Gemini 3.5 Pro — the flagship meant to compete with GPT-5.6 and Claude's latest — missed its June release window and remains "in testing with partners," reportedly over coding performance gaps.
What it means for your business
Notice the pattern: across every major lab, the fast/cheap tier is improving faster than the flagship tier is shipping. For business automation, that's good news — the models that power affordable, high-volume work (chat, triage, data extraction, summaries) keep getting better and cheaper each quarter. Building on the efficient tiers, with the option to escalate hard cases to a flagship, is now the standard cost-smart architecture.
Source: Ars Technica