Google launches Gemini 4 Hydrogen, a faster, pricier Flash successor
For the CTO of a regional bank
Gemini 4 Hydrogen is Google's new model for code, documents and multi-step agents. It costs $1.50 per million input tokens and $7.50 per million output tokens, reasoning tokens included. It scores 51 on Artificial Analysis's index, against 31 for Gemini 3.8 Flash and one point above Gemini 4 Argon, which costs $4/$20 after its introductory pricing ends. The simple model: pay Hydrogen rates and skip the reasoning model. The flaw: Claude Haiku 5.5, GPT-6 Luna and Muse Ember charge $0.10/$0.50, one fifteenth as much. So why not stay on Flash? Flash rises to Hydrogen's exact price on Jan. 1, 2027. The exception is the half-price Contributor tier, which lets Google train on your requests. No loan files go there. Batch, also half price, fits overnight document runs. The model is a price. The routing table is the strategy.
Our take. Hydrogen scores a point above Google's own reasoning model at under half its post-launch price, yet costs fifteen times the lowest standard rates of rival small models. Price which document and agent jobs need that extra point before standardizing.
Action. AI platform engineering: Reprice model-routing budgets for document and agent workloads using Gemini 3.8 Flash's Jan. 1, 2027 rise to $1.50/$7.50.
From our records
- Gemini Enterprise (Google Cloud)vendor · banks using it (11) Wells Fargo, BNY, Citi · alternatives Microsoft Copilot, ChatGPT Enterprise, Claude
- Claude (Anthropic)vendor · banks using it (24) Goldman, JPMorgan, BankUnited · alternatives Microsoft Copilot, ChatGPT Enterprise, Gemini Enterprise
Sources
- ReportGoogle launches Gemini 4 Hydrogen, a faster, pricier Flash successor Venturebeat
- Also reported byGoogle rolls out Gemini 4 Hydrogen, finally replacing 3.8 Flash - Ars Technica Arstechnica
- Also reported byGoogle releases Gemini 4 Hydrogen, a lower-cost model said to outscore Argon Siliconangle