Gemini 4 Carbon

google/gemini-4-carbon

Google reasoning model for sustained agentic software engineering, multimodal knowledge work, scientific reasoning and cybersecurity defense. Generally available in the paid Gemini API and AI Studio. Input, thinking and output share a 1,048,576-token window. Default synchronous output limit is 65,536 tokens; long decode can extend output up to the remaining shared window. Supports implicit caching only. Cursor availability is documented separately.