From synthesizing dense multimodal files to building interactive 3D web experiences, or even uncovering software vulnerabilities, Carbon helps solve problems that require sustained reasoning. Now rolling out in: 🔘 The Google Gemini and AI Mode in Search for Google AI Ultra subscribers. 🔘 Gemini 4 Carbon is available via... the API in Google AI Studio and Gemini for business for all paying developers and enterprises 🔘 Gemini for business, CodeMender, and Google Antigravity for enterprise useshow more

Start building with Gemini 4 Carbon: our strongest model for complex, long-horizon tasks. It’s designed to explore problems thoroughly and run extended verification loops across software engineering, cybersecurity, knowledge work, and science.

With @ArtificialAnlys, we built MKA (Multimodal Knowledge Agent): a benchmark for how well agents handle visually complex tasks. Everyday professional work often uses more than just text – from cross-referencing blueprints to reading scientific charts. This is not covered by existing evaluations. Gemini 4 Carbon ranks #2, at 6x lower average cost per task than the top model. →

@GeminiApp Carbon just took the hardest jobs in the building. Days long debugging, cyber work and research that never sleeps. Argon runs the daily grind, Carbon gets called when it gets serious #Gemini #AI

@GeminiApp Vulnerability hunting and 3D pages in one model is a wide range. I'd start with the boring cases.

@GeminiApp What really matters with long tasks is how well a model handles mistakes along the way, not just perfect benchmark scenarios. Will you share any real examples of Carbon dealing with tool failures or conflicting results?

@GeminiApp Tell me when Gemini can make a PPT like ChatGPT can. I've completely stopped using Gemini lately...thinking about cancelling my subscription...it's just the 5TB I'm enjoying really.

@GeminiApp Interesting

@GeminiApp My first test for Carbon would be hunting software vulnerabilities. I spend my days watching agents fumble long tasks, so a model that holds a thread for hours earns my attention before I trust any launch hype.