By Ken Muthomi - Reporter
USA, July 21 ,2026 - Tech giant Google has introduced two new additions to its Gemini family of artificial intelligence models in a broader effort to provide developers with more efficient AI tools for building production-grade AI agents.
The company said on Tuesday that Gemini 3.6 Flash and Gemini 3.5 Flash-Lite are designed to deliver higher token efficiency, lower latency and more reliable performance, enabling developers to scale agentic workflows while reducing operating costs.
Gemini 3.6 Flash, which now succeeds Gemini 3.5 Flash, delivers improved coding, knowledge work and multimodal capabilities while consuming 17 per cent fewer output tokens, according to the Artificial Analysis Index.
The company says that the model can reduce output token usage by as much as 65 per cent in some benchmarks, lowering the overall cost of running AI-powered applications.
Furthermore, the company said that the model also improves coding by reducing unnecessary code edits and execution loops.
It also enhances machine learning research performance, strengthens computer-use capabilities and offers better document parsing, chart analysis and report drafting for enterprise customers.
Google said that Gemini 3.6 Flash has been equipped with advanced safety measures, especially stronger protections against misuse in chemical, biological, radiological, nuclear (CBRN) and cyber-related domains.
“This enhanced efficiency is also combined with a lower price than 3.5 Flash. At $1.50/1M input tokens and $7.50/1M output tokens, 3.6 Flash reduces the overall cost per agentic task, making agents more cost-effective to build and run,” the company stated.
Google also unveiled Gemini 3.5 Flash- Lite, which it has branded as the fastest and the most affordable model in the Gemini 3.5 series.
The model generates up to 350 output tokens per second and is optimized for high-volume, low-latency tasks such as document processing, agentic search and large-scale AI workflows.
The company said Flash-Lite offers significant improvements over previous Flash-Lite models in coding, long-context understanding and real-world task execution and also includes built-in computer-use capabilities.
Gemini 3.6 Flash and Gemini 3.5 Flash-Lite are currently available through the Gemini API in Google AI Studio, Android Studio and the Gemini Enterprise platform.
Flash-Lite is also available in Google Search. The company said that it is still testing Gemini 3.5 Pro with partners and has already begun pre-training its next-generation Gemini 4 model.
The company has also announced Gemini 3.5 Flash Cyber, a specialised cybersecurity model integrated into its CodeMender security platform.
The model is designed to identify, validate and help fix software vulnerabilities more efficiently by combining multiple AI agents into a single security assessment. It will initially be available only to governments and trusted partners through a limited-access pilot programme.
More from Kenya