Google has launched Gemini 4 Hydrogen, an AI model for coding, debugging and other fast execution tasks. It can work alone or alongside Gemini 4 Carbon and Gemini 4 Argon, which handle longer-horizon reasoning.
Google is making Gemini 4 Hydrogen available to developers through the Gemini API and to paid Google AI Pro and Ultra subscribers through Google Antigravity. Google says the model handles repository coding, debugging, interactive web development and multimodal workflows.
Hydrogen offers 4 configurable thinking levels, with high as the default and xhigh as an additional option. In Antigravity’s Orchestrator, Carbon can explore a codebase and plan changes before Hydrogen makes edits, updates packages and runs tests, handing work back to Carbon if problems arise. Orchestrator with Carbon is rolling out to Google AI Ultra subscribers; Google AI Pro subscribers receive Orchestrator with Argon.
Pricing is $1.50 per million input tokens and $7.50 per million output tokens, flat across the 1 million token context window. For companies building AI agents, this offers a way to separate planning from execution with stated token costs, though Google’s speed claims still need testing against their own workloads.
API access lets developers evaluate Hydrogen in their own agent workflows now.
Handing work between planning and execution models gives builders another way to organise coding agents without assigning every step to the same model.
Configurable thinking levels let builders evaluate different settings within their workflows, without assuming faster completion or lower total costs.
Written automatically by the Made in 10 Newsroom from the source above.
Nebuly has raised $21 million in Series A funding to help enterprises measure the business value of AI agents. Its platform links AI activity to time saved and revenue, rather than simply counting users and prompts.
TSMC’s board has approved plans to invest up to $1 billion in OpenAI and Anthropic. The Anthropic investment would buy common shares at the company’s expected IPO price.
OpenAI has introduced API inference residency in India for eligible businesses using GPT-6 Luna. Supported requests can now be processed within India, alongside local storage of supported customer data.