Google Ships New Gemini Flash Models, But Pro Is Still Missing
Google released three new Gemini models: 3.6 Flash, 3.5 Flash-Lite, and restricted 3.5 Flash Cyber. The launch also highlights a delay: Gemini 3.5 Pro missed its expected release because it reportedly failed internal coding targets, and a training-data update did not fix the issue. Flash models are optimized for speed and lower cost, while Pro models are aimed at deeper reasoning. Gemini 3.6 Flash is the main release. It is cheaper than 3.5 Flash, uses fewer output tokens, and shows stronger benchmark results, especially on software engineering and computer-use tests. It leads OSWorld-Verified and improves sharply on DeepSWE and MLE-Bench, though rivals still outperform it on some coding and knowledge-work benchmarks. Real-world coding tests were mixed, with some outputs needing external fixes to become usable. Gemini 3.5 Flash-Lite targets very high-throughput workloads and is even cheaper, with strong performance for its price. Gemini 3.5 Flash Cyber is limited to governments and vetted partners for vulnerability discovery. Google says work has already begun on Gemini 4.
