Google adds Gemini 4 Hydrogen as trackers log eight AI model releases in a week

Two release trackers say Google shipped Gemini 4 Hydrogen this week, though they disagree on the date. It capped a week that also brought new models from Anthropic, Meta, ElevenLabs and others.

Google adds Gemini 4 Hydrogen as trackers log eight AI model releases in a week

Google has added another model to its Gemini 4 line. Two independent release trackers list Gemini 4 Hydrogen as the newest large language model from a major lab. It is the last release in a week that the trackers say brought eight new models in seven days.

The trackers don’t agree on the date. LLM Gateway, which tracks model availability, says Google released Gemini 4 Hydrogen on 13 November and that the gateway added it the same day. Opper’s model release tracker gives the date as 12 November. Neither listing gives a reason for the gap. The difference may come down to time zones or to when each service first saw the model.

Advertisement

What the listings show

Neither tracker says much about the model. Opper lists Gemini 4 Hydrogen with a context window of 1 million tokens, meaning the amount of text the model can take in at once. The listing also shows two prices, $1.50 and $7.50. Trackers like this one usually show input and output prices per million tokens, so these figures most likely follow that pattern. Google’s own pricing page would be the place to confirm that.

The listings don’t include benchmark results or say what the model is designed for. They also don’t say how Hydrogen fits with the other Gemini 4 models. One of those is Gemini 4 Carbon, which LLM Gateway says Google released on 10 November. Until Google publishes more about the model, buyers comparing it with other Gemini 4 options will have only the tracker data to go on.

Google’s naming also stands out. The company has been giving Gemini 4 variants the names of chemical elements. Opper’s tracker also lists a separate Google product released on 11 November, called Gemini Decision 1. The tracker shows a 262,000-token context window and a single price of $0.10 for that model.

Advertisement

A crowded week

According to LLM Gateway, 11 new AI models had come out in November by the time Hydrogen arrived. Eight of those came in the past seven days. Opper’s tracker shows a similar spread over a slightly different period. The releases the two services list include:

  • Claude Fable 5.2 from Anthropic, dated 9 November. Opper lists it with a 1-million-token context window and prices of $8.00 and $40.00. That makes it far more expensive than Hydrogen on the tracker’s figures.
  • Muse Ember 1.0 from Meta, which LLM Gateway dates to 10 November.
  • Scribe v3 and Scribe v3 Live from ElevenLabs, both dated 10 November.
  • MiniMax H3 Image from MiniMax, also dated 10 November.
  • JT-4.1 Flash 236B A21B from China Mobile, which Opper dates to 11 November.

LLM Gateway also lists a modified DeepSeek model, dated 9 November and labelled “Uncensored.” The label credits the variant to a party called HeimdaLLM, so it appears to be a third-party version and not an official DeepSeek release.

Earlier in the month

The busy week follows several releases in early November. Both trackers say OpenAI released two image models, GPT Image 2.6 Flare and GPT Image 2.6 Sunburst, on 4 November. Opper lists each at $5.00.

Advertisement

Opper’s tracker also shows Tencent’s Hy4 and a model called REAX 1.3 from Decision Models on 3 November. It lists Inkling 2 Small from Thinking Machines on 5 November and Agnes 2.5 Pro from Theta Sapiens AI on 7 November.

Reading the numbers with care

Release trackers are useful for following how fast the industry moves, but they are not official records. They often add a model when it shows up through an API, which may not be the day the company announces it. That may explain the one-day gap in the Hydrogen dates.

Prices and context windows on these sites usually come from the provider’s own documentation, but they can lag behind changes. The trackers also don’t explain every column. Opper shows a number after some prices, for example 54 for Hydrogen and 65 for Claude Fable 5.2, but doesn’t say what it measures in the section reviewed here. For that reason, those figures are not treated here as performance scores.

Advertisement

On the trackers’ figures, the price gap between Google’s and Anthropic’s newest models is wide. Hydrogen’s listed prices are less than a fifth of Claude Fable 5.2’s. Both are listed with the same 1-million-token context window. Whether the two models compete for the same jobs will depend on details that Google and Anthropic have not yet set out in the sources reviewed here.

Sources