Gemini 3.6 Flash release

Google released Gemini 3.6 Flash on July 21, a multimodal, agent-oriented general model accepting text, image, video, audio and PDF input, with a 1,048,576-token context, code execution, function calling, structured output, Search and Maps grounding, and preview computer use. Image and audio generation and the Live API are absent. It is available in the Gemini API, AI Studio and the Gemini app; the live pricing page should be checked before procurement.

Gemini 3.5 Flash-Lite targets high-throughput, lower-cost tasks such as extraction, classification and agent sub-tasks, prioritising efficiency over flagship reasoning depth.

Gemini 3.5 Flash Cyber is a dual-use specialised model that finds, validates and patches software vulnerabilities, in a limited pilot for governments and trusted partners; independent evaluation is needed for real enterprise performance.

NVIDIA's open-source GPU-accelerated medical physics simulation framework in Isaac for Healthcare supports synthetic training, digital twins, reinforcement learning and failure-mode testing; simulation is no substitute for physical, clinical or regulatory validation.