Apple probably uses this model of artificial intelligence in its smart glasses

Apple has taken a big step towards developing smart glasses equipped with artificial intelligence. According to the information released, the company has introduced a new model called Fastvlm that optimally and quickly analyzes high -resolution images. This model can be the basis of artificial intelligence performance in Apple’s future smart glasses.

According to the 9TO5MAC report, there have been many rumors in recent months about Apple’s programs to launch artificial intelligence wearable tools. The company appears to be preparing a version of smart glasses to compete with Ray-Ban Meta glasses that are likely to be released in 2027. Reports also suggest the introduction of cameras with camera and artificial intelligence capabilities. Although it is still too early to judge the final design of these devices, Apple has just taken a brief look at the possible artificial intelligence of these products.

Fastvlm’s artificial intelligence model can be the brain of Apple’s smart glasses

The year 2023, the Apple Machine Learning Research Team introduced a superficial and lightweight MLX framework, designed to teach and execute machine learning models locally on devices equipped with Apple Silicon. This framework is also compatible with the common languages ​​and tools of the development of artificial intelligence models.

Apple has now introduced a model called Fastvlm that is the Visual Language Model using MLX and can process images at high resolution and speed. This model is considerable and faster than similar models.

At the heart of Fastvlm, there is a refinery called “Fastvithd” designed for effective processing of high -resolution images. According to Apple, the model is up to 3.2 times faster and 3.6 times smaller than similar models. This is very important; Because for devices such as smart glasses that need to process information locally, high productivity and low consumption are essential.

Fastvlm also uses less tokens to produce output, which makes the Inference step faster. According to Apple, their model has an initial response time 85 times faster than similar models. This model is published on the Gate Hall and its scientific article is available in Arxiv.

The Fastvlm model shows that Apple is working seriously on the development of native artificial intelligence infrastructure for its future products, such as smart glasses. By reducing resource consumption and increasing processing speed, this model can provide a smooth and without dependence on cloud space; This feature is crucial to the next -generation wearable tools.