Gemini 4 Hydrogen now available on AI Gateway

Vercel's AI Gateway now serves Google's Gemini 4 Hydrogen, the fast, lower-cost model in the Gemini 4 family that Google released on November 12, the same day as this changelog. The Gateway is a single API that sits in front of many model providers: a developer authenticates once with Vercel, names a model in each request, and the Gateway forwards the call, tracks usage and cost, and lets the developer configure retries and failover. For anyone already sending traffic through it, this changelog means Hydrogen is reachable by setting the model name to google/gemini-4-hydrogen. Vercel's announcement shows that model name working in its AI SDK, in an OpenAI-compatible Chat Completions endpoint, and in coding agents such as Claude Code, Codex and Cursor after running Vercel's CLI setup command, npx vercel ai-gateway setup. The capability claims in the post (better agentic coding, tool use and grounding than Gemini 3.8 Flash in the same latency class) come with no benchmark figures or independent testing.

On cost, the changelog makes one promise: Vercel says the Gateway passes through provider pricing with no markup and charges no platform fee on inference, including when customers bring their own Google key. The post itself lists no prices, but Vercel's model page shows $1.50 per million input tokens and $7.50 per million output tokens with a 1-million-token context window, matching Google's list price, and names both Google's direct API and Vertex AI as serving providers. Neither page states rate limits. Bringing your own key has a catch that Vercel's BYOK documentation spells out: if a request made with your key fails, the Gateway retries it on Vercel's own credentials and bills that fallback to your Gateway credits, and BYOK requires a paid tier with purchased credits. The free tier Google offers for Hydrogen on its own Gemini API is not mentioned for the Gateway, so developers who want free usage should not assume it carries over.