Google expands Gemini with cheaper models and a bug-hunter it keeps on a leash - SiliconANGLE

Google today unveiled three new Gemini Flash models—Gemini 3.6 Flash, Gemini 3.5 Flash‑Lite, and Gemini 3.5 Flash Cyber—designed to lower AI costs while boosting performance. Gemini 3.6 Flash, the flagship, cuts up to 17 % fewer output tokens than its predecessor, is priced at $1.50 per million input tokens and $7.50 per million output tokens, and shows significant gains on coding, knowledge, and machine‑learning benchmarks. Gemini 3.5 Flash‑Lite is the fastest and cheapest option, priced at 30 cents per million input tokens and $2.50 per million output tokens, and is positioned as a migration path for workloads that previously ran on Gemini 2.5 and 3 Flash. Gemini 3.5 Flash Cyber is a security‑tuned model that runs inside CodeMender and is restricted to governments and trusted partners. CodeMender, a previewed code‑security agent built on DeepMind research, scans repositories for vulnerabilities, verifies them by building and running exploits in a customer‑controlled sandbox, and then writes patches for developer approval. It supports C/C++, Go, Java, Python, Ruby, Rust, and TypeScript, and keeps a human in the loop to approve patches before committing. Early enterprise testers—including Salesforce, Robinhood, and Palo Alto Networks—reported that CodeMender accelerated the path from vulnerability detection to tested fixes and identified critical issues that other AI tools missed. The new Gemini models and CodeMender are available through the Gemini API, AI Studio, Android Studio, and the Gemini Enterprise Agent Platform, with plans to add third‑party frontier models later this year.

(Source：Siliconangle)