Industry News
The fourth Gemini in a month: Gemini 4 Neon priced at $3 and $15, 1 million contexts and 1 million outputs
Google has released a new model called Gemini 4 Neon. This is the fourth Gemini model in a month, following the Argon, Carbon and Hydrogen.
Its location is sandwiched in the middle. Hydrogen is cheap and fast, Carbon is the strongest but slow, and Neon needs to have both. Google says it's 3 times faster than Carbon and 40% cheaper to complete the same task.
In terms of running points, it does squeeze into the first echelon. In high-intensity thinking mode, Neon scored 61 points in the Artificial Analysis Intelligence Index, ranking third in the overall list, preceded only by larger and slower models; it ranked second in the Vals Index, and was on the Pareto frontier of cost-effectiveness.
A relatively rare parameter is the output length: the context is about 1 million tokens, and a maximum of 1 million tokens of output can be generated at a time. The output upper limit of most models is only a fraction of the context. This design is more friendly to scenarios where long code bases and long reports are generated at one time.
The API is open today, with pricing starting at $3 per million inputs and $15 per million outputs.
This price should be viewed on the timeline. Carbon is currently listed at promotional prices of US$2 and US$10, which is cheaper than Neon; the promotional price expires on November 30, and will return to US$4 and US$20 from December 1, when Neon is a quarter cheaper. The conclusion drawn from the price comparison today will be reversed next month.
As for "40% cheaper", this is the general ledger calculated on a per-task basis, provided that Neon thinks less than Carbon. The unit price is more expensive. If your task does not require long thinking, the money saved may not be as much as advertised. It is best to run your own task again before changing.