💎Google Ships Gemini 4 Carbon at $2 Per Million Input Tokens

Your coding agent just got cheaper and smarter

Google just dropped Gemini 4 Carbon, and the pitch is simple: frontier accuracy at prices that don't make your finance team flinch. It ranks #1 on CursorBench 4.0, Perplexity's WANDR, and Zapier's AutomationBench, all at more than 2X lower cost per task than the next best external model. If you're building agents, this is the part to care about. Carbon is aimed at multi-step coding, debugging, and long-horizon agentic workflows, and it recovers from its own errors by running build-and-verify loops. It handles multi-package migrations and cross-subsystem root-cause investigations, which is where most models fall over. It's available through the API in Google AI Studio and Gemini for business. The numbers: introductory pricing is $2 per million input tokens and $10 per million output tokens. On the Multimodal Knowledge Agent benchmark it lands at #2, at more than 6X lower average cost per task than the top performer. Google AI Ultra subscribers also get it in Antigravity, the Gemini app, and AI Mode in Search. The catch? Introductory pricing is exactly that, introductory. Lock in your cost assumptions before you rebuild anything around it.