Overview
- The rollout began Tuesday, November 10, and makes Gemini 4 Carbon available to paid Gemini API developers, enterprises, Google AI Ultra subscribers in the Gemini app and AI Search mode, and through Google Antigravity.
- Google positions Carbon as its top‑tier model for the hardest tasks such as multi‑step coding, cross‑subsystem debugging, and open‑ended scientific work while keeping Argon as the recommended model for most engineering and knowledge tasks.
- Antigravity’s previewed /swarm turns a single prompt into a coordinated team of agents so teams can assign Carbon to deep reasoning roles and run lighter models in parallel to write, verify, or audit changes in real time.
- Google launched Carbon with aggressive introductory pricing of $2 per million input tokens and $10 per million output tokens until November 30, after which those rates double, and the company says Carbon can use more tokens per task because it checks and verifies outputs more thoroughly.
- Google cites top rankings on new and existing benchmarks and offers examples where Carbon helped formally verify Signal code and where security partners found zero‑day flaws quickly, but independent third‑party validation, broader free access and real‑world cost impacts remain unclear.