Google Releases Gemini 4 Carbon AI Model

Google officially released its latest AI model, Gemini 4 Carbon, to general developers, enterprise customers, and paid subscribers on November 10. The model is now available through Google AI Studio and Android Studio for developers, via Gemini for Business for enterprises, and through the Google AI Ultra plan for general users.

According to aitimes.com, the company is positioning the model as a cost-effective alternative to competitors. API pricing is set at $2 per million input tokens and $10 per million output tokens until November 30, after which rates will increase to $4 and $20, respectively. In comparison, Anthropic’s Claude Fable 5.2, which launched on November 9, costs $8 per million input tokens and $40 per million output tokens.

Google reports that Gemini 4 Carbon outperformed competing models on several industry benchmarks, including CursorBench 4.0, Perplexity's WANDR, and Zapier's AutomationBench, while operating at less than half the cost per task. On AutomationBench, the model achieved a 53.88% success rate at $2.12 per task, compared to the 51.45% success rate of Claude Fable 5.2 at $4.64. Additionally, Google previewed a /swarm feature for its Antigravity coding platform, allowing multiple agents to collaborate on prompts.

The release follows the initial unveiling of Gemini 4 Carbon on October 28, when access was restricted to cybersecurity experts through the Fairwind program. While the model briefly led the Artificial Analysis intelligence index upon its October debut, it was subsequently surpassed by Claude Fable 5.2. Current rankings on the index place Claude Fable 5.2 at 65, followed by Gemini 4 Carbon at 60, Claude Opus 5.5 at 59, and GPT-6 Astra at 57.

Google also expanded its researcher support program, providing selected scientists with six months of free access to both Gemini 4 Carbon and the Argon model. Meanwhile, the specialized Carbon for Cyber remains restricted to approved security professionals.