Google has released Gemini 4 Carbon, its strongest AI model yet, to developers, enterprises, scientists, and consumers. Carbon is designed for frontier-level tasks including cybersecurity, long-horizon agentic workflows, complex knowledge work, and scientific reasoning. It builds on the recently launched Gemini 4 Argon, pushing the frontier further for multi-step coding, debugging, and open-ended research.
In independent benchmarks, Carbon ranks #1 on Cursor Bench 4.0, Perplexity’s WANDR, and Zapier’s Automation Bench, delivering frontier accuracy at more than 2X lower cost per task than the next best external model. It also excels in multimodal agentic workflows, ranking #2 on the new Multimodal Knowledge Agent benchmark at over 6X lower average cost. In pre-release evaluations, it matched top models on deep financial research and ranked first out of 23 frontier configurations on financial intelligence.
Gemini 4 Carbon is available now via the API in Google AI Studio and Gemini for Business at $2 per million input tokens and $10 per million output tokens. It is also rolling out to Google AI Ultra subscribers in Google Antigravity, the Gemini app, and AI Mode in Search. Google is previewing /swarm, a feature that turns a single prompt into a coordinated team of agents.
Source: Hacker News · Summarized by HeadlinesBriefing