12 November 2026 — Google has opened Gemini 4 Carbon, its strongest AI model yet, to paying developers,
businesses and AI Ultra subscribers — and unveiled a new robotics model alongside it.
Google said on 11 November it will broaden access to Gemini 4 Carbon, which was released on 28 October to some cybersecurity personnel and partners, to developers, businesses, researchers and the general public. The model is built for complex, long-running work such as coding, cybersecurity and scientific research.
What Gemini 4 Carbon is
- Positioning: Google’s top-tier model, designed for long agent jobs — coding, research and security work that runs for hours rather than minutes.
- Pricing: Developers pay $2 per million input tokens and $10 per million output tokens until 30 November. The price then doubles to $4 and $20.
- Where it is available: In the Gemini app and AI Mode in Search for Google AI Ultra subscribers, through the Gemini API in AI Studio and Android Studio for paying developers, in Google Antigravity for Ultra subscribers, and in Gemini for business and CodeMender.
- Early results: Cursor says Carbon at high effort is the top model on its CursorBench at 60.4%. Perplexity has rolled it out to Pro and Max subscribers and says it is 2.5 times cheaper than Anthropic’s Fable 5.2 with similar results on agentic research tasks.
Gemini Robotics 2.5
Unveiled alongside Carbon, Gemini Robotics 2.5 is Google DeepMind’s updated vision-language-action model for robot manipulation. It is rolling out to robotics partners, and Google reports a 96% success rate in an Apollo 2 automotive-kitting proof of concept. DeepMind says the model can learn hundreds of everyday skills, from folding laundry and serving pastries to sorting car parts.
Why it matters
The launch caps a remarkable 24 hours in AI: Google, Anthropic and OpenAI each shipped major tools within a day of each other. Anthropic launched Claude Meetings, Sheets and Whiteboards in beta, while OpenAI launched Businesses in ChatGPT in beta. On the MKA benchmark cited by The Daily Fledgling, Carbon scored 54%, just behind Anthropic’s Fable 5.2 at 57% — but at a far lower cost per task.
For users, the practical change is simple: the model Google previously kept for security specialists and partners is now the one answering in the Gemini app and API.
Related on fetlife.news: Anthropic Launches Claude Fable 5.2 and Mythos 5.2: 30% Faster AI Models — Tech; Anthropic Sets IPO Price at $70 to $78, Valuing Claude Maker at Up to $1.4 Trillion — Tech
External references: ChosunBiz, 11 Nov 2026; The Neuron, 11 Nov 2026; SXF Brief, 11 Nov 2026.