SAN FRANCISCO, Feb 5 - Google, a subsidiary of Alphabet, announced updates to its Gemini series of large language models, unveiling a new product line aimed at competing with low-cost artificial intelligence models like those from Chinese competitor DeepSeek.
The technology company provides various versions of Gemini with differing price points and performance levels. It previously introduced a lightweight option called Flash, and now has introduced an even more affordable option called Flash-Lite.
Google made Gemini 2.0 Flash available to the general public on Wednesday, following its release to developers in December. Alongside this release, the company also introduced Flash-Lite and initiated testing of an updated version of its flagship "Pro" model.
Chief Technology Officer of Google's DeepMind AI lab, Koray Kavukcuoglu, mentioned in a press release that positive feedback on Flash version 1.5 prompted the development of Flash-Lite. Notably, Gemini 2.0 Flash comes with a higher price tag compared to its predecessor.
Recent weeks have seen increased scrutiny on the cost of developing and utilizing AI models, especially after DeepSeek disclosed spending under $6 million on the final training phase of a model. This revelation sparked discussions during earnings calls for Alphabet and its competitors, all of which have indicated continued significant investments in the AI sector.
Alphabet shares saw a decline on Tuesday, partially influenced by news of a planned capital expenditure increase that exceeded Wall Street's expectations by 29%.
Gemini Flash-Lite is priced at $0.019 per 1 million tokens, a unit of data processed by an AI model. This is in contrast to OpenAI's flagship model at $0.075 per million tokens and DeepSeek's low-cost model at $0.014 per million tokens. However, DeepSeek has announced plans to raise its prices fivefold on Feb. 8.