San Francisco:
Alphabet’s Google on Wednesday announced an update to its Gemini family of the big language model, including a new product line with competitive pricing for low -cost artificial intelligence models such as Chinese rival Deepsek.
Tech giants offer several versions of Gemini that differ in value and performance. It already offers a light version known as “flash”, but its new “flash-light” model is even more cheaper.
On Wednesday, Google released Gemini 2.0 flash to the general public after previewing developers in December. It also launched the flash-light and released a new version of its major “Pro” model in the test stages.
Google made flash-light after receiving positive response to the 1.5 version of the flash, Kore Kavukuoglu, Chief Technology Officer of Deepmind AI Lab, Google’s Deepmind said in a press release. Gemini 2.0 flash is expensive than its predecessor.
The cost of developing AI models and in turn, the cost of using them has come under investor investigation in recent weeks, when Deepsek revealed that it had spent less than $ 6 million on a model’s final training run run. Developers of major US AI firms said that the total cost was more likely.
Nevertheless, Uday of Dipsek raised questions on the earnings call of the alphabet and rivals Microsoft and Meta. All have so far indicated with the intention of continuing large -scale capital expenditure in the region.
The alphabet’s shares slipped into the part on Tuesday due to the investors pessimism around a planned Capex Hike, which was 29% higher than the expectation of Wall Street.
A word for units of data processed by AI model, costing some input $ 0.019 per 1 million tokens on Gemini flash-light. This compars $ 0.075 on the cost-skilled version of the major model of openiAI and $ 0.014 on the cheap model of Deepsek, although Deepsek stated on its website that pricing will increase five times on February 8.
(Tagstotransite) Google (T) AI (T) Deepsek
