Google introduces new class of cheap AI models as cost concerns intensify
Google has announced updates to its Gemini language models, introducing a new "Flash-Lite" product that is competitively priced against low-cost AI models, such as those by Chinese competitor DeepSeek. This move comes as part of Google's strategy to diversify its offerings with various versions of Gemini that differ in cost and performance. The company has released Gemini 2.0 Flash to the public, after a developer preview, and introduced Flash-Lite alongside a new version of its "Pro" model for testing.
These updates follow positive reception of the previous Flash model, prompting Google to develop the more affordable Flash-Lite, according to Koray Kavukcuoglu, CTO of Google's DeepMind AI lab. The decision to offer cost-effective AI solutions reflects growing investor scrutiny over the development expenses of AI models. This scrutiny intensified after DeepSeek disclosed its relatively low expenditure of under $6 million on its model's final training run, highlighting the stark contrast with the presumably higher costs faced by leading US AI developers. Google's latest releases aim to maintain competitiveness in the evolving AI landscape while addressing cost concerns.
via SCMP Full Text Feed
Google has announced updates to its Gemini language models, introducing a new "Flash-Lite" product that is competitively priced against low-cost AI models, such as those by Chinese competitor DeepSeek. This move comes as part of Google's strategy to diversify its offerings with various versions of Gemini that differ in cost and performance. The company has released Gemini 2.0 Flash to the public, after a developer preview, and introduced Flash-Lite alongside a new version of its "Pro" model for testing.
These updates follow positive reception of the previous Flash model, prompting Google to develop the more affordable Flash-Lite, according to Koray Kavukcuoglu, CTO of Google's DeepMind AI lab. The decision to offer cost-effective AI solutions reflects growing investor scrutiny over the development expenses of AI models. This scrutiny intensified after DeepSeek disclosed its relatively low expenditure of under $6 million on its model's final training run, highlighting the stark contrast with the presumably higher costs faced by leading US AI developers. Google's latest releases aim to maintain competitiveness in the evolving AI landscape while addressing cost concerns.
via SCMP Full Text Feed
