Google launches low-cost AI models for image and video: competition with cheap alternatives

Google has launched two new low-cost artificial intelligence models aimed at image and video processing, expanding its Gemini Enterprise Agent platform. The recently announced initiative includes the Gemini 3.1 model Flash-Lite Image, designed to offer competitive performance with reduced prices, targeting companies that seek economic alternatives to traditional options. The move reinforces the fierce dispute in the business AI market, where giants like Amazon and Microsoft also invest in lighter and more affordable solutions. For more details, see MediaPost.

Cheap AI models: How does Google reduce costs?

Google's bet is on creating smaller and more efficient architectures that require less computational power to perform specific tasks. The Gemini 3.1 model Flash-Lite Image is an example: it eliminates unnecessary layers of processing in images, maintaining precision in object recognition and visual classification. This reduces inference costs on cloud servers, allowing companies to pay less per use. The strategy is especially relevant for startups and SMEs that need quality AI without compromising the budget.

Meet the new cheap Google AI models

In addition to the image-focused model, Google also introduced a specialized video version, Gemini 3.1 Flash-Lite Video, which analyzes frames optimally for tasks such as content moderation and transcription. Both run on the Gemini Enterprise Agent platform, which offers integration tools with REST APIs and SDKs. The Gemini 3.1 Flash-Lite Image operates with reduced parameters compared to the Gemini 3.1 standard model, but maintains high performance for specific usage scenarios such as e-commerce catalogs or scanned document analysis.

Cheap AI models: impact for Brazilian companies

In Brazil, where the cost of cloud infrastructure still weighs on the IT budget, cheaper AI models can accelerate the adoption of smart solutions. Retail, logistics and health companies, for example, can use technology to automate quality inspection or product recognition without relying on expensive hardware. The tendency of intense adoption of AI is already increasing hiring in technical areas, as shown in a recent study: Intense adoption of AI increases hiring, reveals Ramp study.

Competition with other cheap AI models

Google's movement is not isolated. Companies like Meta and OpenAI already offer low-cost satellite models, and the arrival of Gemini 3.1 Flash-Lite Image closes the competition. Google's main difference is native integration with the Gemini ecosystem, which includes corporate security and compliance tools. For corporations that already use Google Cloud, migration can be smoother. The exact price per API call has not yet been released, but the expectation is that it is up to 40% lower than standard models of competition, according to industry sources.

Frequently Asked Questions about Cheap AI Models

How much are Google's new AI models?

Official prices have not yet been published, but it is estimated that the cost per API call is up to 40% lower than traditional models of competition, focusing on competing with existing low-cost solutions.

Are the models available in Brazil?

Yes, the models are available globally via the Gemini Enterprise Agent platform, accessible in clouds from Google Cloud in the South American region. Brazilian companies can test via API.

What does an AI Lite model mean?

'Lite' models are optimized versions of larger architectures with fewer parameters and less computational consumption. They sacrifice a little accuracy in general tasks for efficiency and low cost, being ideal for specific applications such as real-time image analysis.