Google announced Gemini 4 Hydrogen today, the least expensive member of the Gemini 4 family. Hydrogen is introduced as the new free-tier model and will power AI features in Google Search. Google positions Hydrogen as an execution-focused model for everyday work: writing code, running tests, and verifying changes quickly and efficiently, and as a companion execution engine to the higher-capability models Gemini 4 Carbon and Gemini 4 Argon in multi-model workflows.
Pricing — figures and dates
Google's current pricing scheme is presented as follows (USD):
-
Gemini 4 Carbon
- Through Nov 30, 2026: Input $2.00, Cached input $0.10, Output $10.00
- From Dec 1, 2026: Input $4.00, Cached input $0.20, Output $20.00
-
Gemini 4 Argon
- Through Nov 30, 2026: Input $2.00, Cached input $0.10, Output $10.00
- From Dec 1, 2026: Input $4.00, Cached input $0.20, Output $20.00
-
Gemini 4 Hydrogen (current price)
- Input $1.50, Cached input $0.075, Output $7.50
Carbon and Argon are priced the same today and both are scheduled to double in price on December 1, 2026. At present, Hydrogen's pricing is about 25% lower than Carbon and Argon; after the December 1 price increase for Carbon and Argon, Hydrogen would represent roughly a 75% discount relative to those models.
Competitors and relative cost
Hydrogen's price remains substantially higher than some competing models mentioned in the report: Claude Haiku 5.5 and GPT-6 Luna are cited at $0.10/$0.50 (in the same input/output pricing terms referenced), which makes Hydrogen roughly 15 times more expensive than those alternatives.
Performance positioning and model roles
Google explains that Argon “delivers Gemini 4 intelligence with faster responses and lower token spend per turn compared to Gemini 4 Carbon.” Although Carbon and Argon share the same price now, Argon may use fewer tokens per task and respond faster, which could justify both models existing at parity pricing.
Hydrogen is intended as the high-efficiency execution engine for routine tasks, while Carbon and Argon serve as frontier models for deeper architectural judgment and complex debugging. Hydrogen will also serve as the free-tier option and the model powering Search AI features.
Quality observations
In demonstrations, Hydrogen was used to generate images of pelicans, and a published comparison grid shows outputs for all three Gemini 4 models across different effort levels. Given the relatively small price gap among the three models today, the differences in pelican output quality for Hydrogen were described as quite stark compared with its larger siblings.
Conclusion
Gemini 4 Hydrogen is positioned as a cost-conscious, execution-focused model within the Gemini 4 lineup and as the new free-tier model for Google. Currently it is somewhat cheaper than Gemini 4 Carbon and Argon, but not dramatically so; after the scheduled December 1, 2026 price increases for Carbon and Argon, Hydrogen will represent a much larger discount. Hydrogen nonetheless remains more expensive than some competing offerings, and differences in token efficiency and output quality among the Gemini 4 models will likely influence developers' and companies' model choices for daily workflows and cost optimization.
