Google DeepMind released three new Gemini models on Tuesday, 2026-07-21: Gemini 3.6 Flash, Gemini 3.5 Flash-Lite, and Gemini 3.5 Flash Cyber. For CTOs and AI product teams, the release expands Google’s Flash lineup while leaving its highest-capability Pro model unavailable for broad use.
What changed
TechCrunch and CNBC reported that Google released Gemini 3.6 Flash, Gemini 3.5 Flash-Lite, and Gemini 3.5 Flash Cyber on Tuesday. According to Google, the release is aimed at efficiency, latency, and reliability for customers building AI agents at scale.
Each model has a distinct role. TechCrunch reported that Gemini 3.6 Flash is positioned for improved coding, knowledge work, and multimodal performance. CNBC reported that Gemini 3.5 Flash-Lite is Google’s fastest and least expensive model in the 3.5 family, while PCWorld described it as optimized for low-latency work such as agentic search. TechCrunch and CNBC reported that Gemini 3.5 Flash Cyber is tuned for cybersecurity work, including finding, detecting, and patching software vulnerabilities.
Equally notable is what did not launch. TechCrunch reported that this update did not include Gemini Pro, Google’s flagship model. Google said Gemini 3.5 Pro is being tested with partners ahead of broader availability.
Why B2B teams should care
The clearest operational takeaway in this Google Gemini models release is efficiency. TechCrunch and CNBC reported that Gemini 3.6 Flash uses up to 17% fewer tokens than the prior model. That is not the same as a guaranteed 17% bill reduction, because actual savings depend on prompt shape, response length, routing, and fallback behavior.
Several secondary reports, including TradingKey and Yahoo Tech, reported pricing of $1.50 per million input tokens and $7.50 per million output tokens for Gemini 3.6 Flash, and $0.30 per million input tokens and $2.50 per million output tokens for Gemini 3.5 Flash-Lite. An official Google pricing page is not included in the source set for this review.
CNBC separately reported that Gemini 3.6 Flash costs less per token than the previous model, and that Flash Cyber is positioned by Google at a lower price per token than larger cybersecurity models.
Who is affected
Based on the announced positioning, three buyer groups are most likely to evaluate these models.
First, teams building AI agents at scale are the clearest target, because Google explicitly framed the launch around efficiency, latency, and reliability for that use case.
Second, teams that need low-latency, high-throughput inference will likely focus on Gemini 3.5 Flash-Lite. PCWorld reported that Gemini 3.6 Flash and Gemini 3.5 Flash-Lite are available now in the Gemini app. TradingKey additionally reported that both are available through the Gemini API, Google AI Studio, and Gemini Enterprise. TradingKey also reported plans for Flash-Lite to appear in Google Search.
Third, security teams and some public-sector buyers may evaluate Gemini 3.5 Flash Cyber. Google said, as reported by TechCrunch and CNBC, that Flash Cyber will be available only to governments and trusted partners in a limited-access pilot.
What teams should check now
Teams evaluating Gemini should verify which current workloads matter most, because Google segmented this release by use case rather than around a new Pro model:
- coding, knowledge work, or multimodal tasks that map to Gemini 3.6 Flash
- lowest-latency tasks such as agentic search that map to Gemini 3.5 Flash-Lite
- security-specific vulnerability detection and remediation that map to Gemini 3.5 Flash Cyber
Engineering and FinOps teams should also review model-routing and budget assumptions. The reported 17% token-efficiency change for Gemini 3.6 Flash and the much lower reported per-token pricing for Flash-Lite are both large enough to change which model should handle first-pass inference, retries, or background agent steps.
Security teams and regulated buyers should confirm whether they qualify for the Flash Cyber limited pilot before planning around it.
What remains unclear
- Not yet confirmed: an official broad launch date for Gemini 3.5 Pro.
- Not yet confirmed: exact official Google pricing documentation for Gemini 3.6 Flash, Gemini 3.5 Flash-Lite, or Gemini 3.5 Flash Cyber.
- Not yet confirmed: complete benchmark tables or context-window details for the new models.
- Not yet confirmed: whether Bloomberg’s reported performance-related delays reflect Google’s direct explanation, rather than external reporting cited by other outlets.
What to watch next
TechCrunch reported that Gemini Pro was last updated in February, and that Google teased Pro in May while saying it was already being used internally. Google’s current position, as reported by TechCrunch and CNBC, is that broader 3.5 Pro availability is still coming.
TechCrunch and CNBC reported that Google has begun its largest-ever pretraining run for Gemini 4.
Until Gemini 3.5 Pro reaches broad availability, Google’s enterprise portfolio remains centered on specialized Flash models rather than a new flagship offering.
Sources
- TechCrunch, Google releases three new Gemini models — but no 3.5 Pro
- CNBC, Google Gemini Flash AI mythos rival
- PCWorld, Google has new Gemini models, but the one everyone wants isn’t here
- TradingKey, Google Strikes Before Q2 Earnings: Releases Gemini 3.6 Flash and Three New Models Focused on Extreme Cost-Performance
- Yahoo Tech, Google launches 3 Gemini AI models
