Google ships three new Gemini Flash models but its frontier 3.5 Pro remains lost in training
Quick Answer
Google has launched three new Gemini Flash models, including 3.6 Flash and 3.5 Flash Cyber, while the anticipated Gemini 3.5 Pro remains in private testing.
Quick Take
The 3.6 Flash model offers significant efficiency improvements and lower costs, outperforming its predecessor in benchmarks, but Google lags behind competitors in the frontier model space.
Key Points
- Gemini 3.6 Flash reduces token usage by 17% compared to 3.5 Flash.
- 3.5 Flash-Lite achieves 350 output tokens per second at lower costs.
- Gemini 3.5 Flash Cyber scores 83.2% on CyberGym, close to OpenAI's GPT-5.5-Cyber.
- Google's flagship Gemini 3.5 Pro is months behind schedule and still in private testing.
- Competitors like OpenAI and Anthropic have already released advanced models.
DeepSignal Analysis
What happened
Google has introduced three new models in the Gemini Flash series: 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber. However, the anticipated Gemini 3.5 Pro remains in private testing and is not yet available to the public.
Key evidence
- The Gemini 3.6 Flash model is reported to use about 17% fewer output tokens than its predecessor, 3.5 Flash, and offers significant cost reductions.
- Gemini 3.5 Flash Cyber, designed for cybersecurity tasks, scored 83.2% on the CyberGym benchmark, closely trailing OpenAI's GPT-5.5-Cyber.
- Google's flagship model, Gemini 3.5 Pro, is still in private testing, leaving the company without a competitive offering in the frontier model space.
Why it matters
The launch of the new Gemini Flash models indicates Google's focus on efficiency and cost reduction, but the absence of the 3.5 Pro model suggests a gap in their competitive positioning. As rivals like OpenAI and Anthropic advance with their frontier models, Google's delay could impact its market share and innovation perception.
What to watch
Source Excerpt
Google is shipping three new Flash models in the Gemini series, including the more efficient 3. 6 Flash, which uses up to 65 percent fewer tokens, and a cybersecurity model available only to governments and select partners. But the anticipated flagship, Gemini 3. 5 Pro, is still missing, while OpenAI, Anthropic, and Chinese labs are already competing at the frontier level.
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from The Decoder
See more →
An AI model programmed nonstop for 19 days on a single MirrorCode task that cost $2,600 to run
Epoch AI's MirrorCode benchmark reveals Claude Opus 4.7 as the leader with a 56% solve rate, reconstructing a 16,000-line toolkit in 14 hours. Despite this, all models tested struggle with the most complex tasks, highlighting limitations in current AI capabilities. The single task consumed $2,600 over 19 days, raising questions about cost-effectiveness in AI development.
