[NVIDIA launches new model and agent tool support for local AI speech model RTX Spark] On October 28th, NVIDIA announced that the Nemotron 3.5 Transcribe open weight speech model has supported RTX Spark, which can realize speech transcription, speaker recognition and translation, and supports 40 languages. Liquid AI’s LFM3 family of models has also gained support and can be run locally and fine-tuned. In addition, NVIDIA has added a new GPU resource management function, allowing users to allocate GPU computing power between foreground applications and background AI agents. RTX Spark notebooks are currently launched by manufacturers such as ASUS, Dell, HP, Lenovo, Microsoft and MSI, starting at US$2,599, equipped with up to 128GB of unified memory, and FP4 AI computing power up to 1 petaflop.