Google launches Gemini 4 Hydrogen as its new low-cost AI model

The new Gemini 4 lineup

Google has officially launched Gemini 4 Hydrogen, which serves as the most affordable option within the new Gemini 4 family. This release follows the introduction of two larger models, Carbon and Argon, which are designed for more complex reasoning tasks. Hydrogen is specifically positioned to handle everyday work and high-speed execution.

The pricing for the new family is somewhat complex and varies based on the date of use. Currently, Carbon and Argon are priced at $2.00 per million input tokens, while Hydrogen sits slightly lower at $1.50. These prices are set to change significantly starting December 1, 2026, when the larger models are scheduled to double in cost.

As simonwillison.net reported, the current price difference between Hydrogen and its more capable peers is only about 25%. However, once the price hike for Carbon and Argon takes effect next year, Hydrogen will represent a 75% discount compared to those models. This suggests Google is preparing a tiered ecosystem where Hydrogen becomes the clear budget choice.

Distinguishing the models

A common question for users is why Google offers two models, Carbon and Argon, at the exact same price point. Google explains that Argon is designed to deliver the same level of intelligence as Carbon but with faster response times. It is also intended to use fewer tokens per turn, making it more efficient for certain types of interactions.

Hydrogen occupies a different niche entirely by focusing on disciplined execution rather than deep architectural judgment. While teams might use Carbon or Argon to untangle subtle bugs or design complex systems, they can turn to Hydrogen for repetitive tasks. This includes activities like writing basic code, running automated tests, and verifying specific changes.

The model is built to function as a standalone tool for routine work. It can also be integrated into multi-model workflows where it acts as the high-efficiency engine. By separating 'thinking' from 'doing,' Google allows users to optimize their costs based on the difficulty of the task at hand.

Integration with Google Search

One of the most significant updates regarding Hydrogen is its role in the broader Google ecosystem. The company has designated Hydrogen as the new free-tier model for its services. This means it will be the primary engine powering AI features within Google Search for the general public.

By using a lower-cost model for search, Google can provide AI-generated summaries and features to a massive audience without incurring the high costs of frontier models. This strategy allows the company to maintain a free service while still leveraging the capabilities of the Gemini 4 architecture. It represents a move toward making AI a standard, integrated part of the web experience.

For the average user, this means the AI interactions they see in search results will be powered by Hydrogen. While it may not have the same depth as Carbon, it is optimized for the speed and reliability required for search queries. It serves as a gateway for millions of people to interact with Google's latest AI technology daily.

Market competition and costs

Despite being the 'cheap' model in the Gemini 4 family, Hydrogen is still notably more expensive than some competitors. For example, Claude Haiku 5.5 and GPT-6 Luna are both priced at $0.10 for input and $0.50 for output. This makes Hydrogen approximately 15 times more expensive than those specific rival models.

The pricing structure highlights a gap between Google's current offerings and the ultra-low-cost models available in the market. While Hydrogen is a significant step for Google's free-tier strategy, it remains in a higher price bracket than the most economical options from other providers. This may influence how developers choose which models to use for high-volume, low-complexity tasks.

The upcoming price changes in December 2026 will be a key moment to watch for the market. If the larger models double in price while Hydrogen stays steady, it will create a much clearer distinction in the hierarchy. This shift will likely define how businesses allocate their budgets between different tiers of AI intelligence.