NVIDIA adds Nemotron 3.5 Transcribe to RTX Spark laptops

By Infinithink Studio · October 28, 2026

NVIDIA says RTX Spark laptops are now available and introduced Nemotron 3.5 Transcribe, an open weight speech model for transcription, speaker labels and translation. For small teams the question is where recordings are processed and who checks the text.

What happened

According to NVIDIA, in a post on its official blog dated October 28, 2026, RTX Spark laptops are now available from ASUS, Dell, HP, Lenovo, Microsoft and MSI. NVIDIA says these systems combine an NVIDIA Blackwell RTX GPU and an NVIDIA Grace CPU in a single superchip, with up to 128GB of unified memory.

The company also introduced NVIDIA Nemotron 3.5 Transcribe, a new open weight speech model. According to NVIDIA, one promptable model handles transcription, speaker labels and translation, and can also answer questions about the audio. NVIDIA says it transcribes 40 languages.

Developers guide the model with plain-language prompts. NVIDIA gives two examples, asking it to transcribe a recording and label each speaker, or to list the decisions made in a meeting. Prompts can also include product terms and formatting conventions, which NVIDIA says helps with specialized vocabulary in fields like healthcare, financial services and customer service.

NVIDIA states that the model ranks No. 1 on the Hugging Face Open ASR and No. 2 among open weight models on Artificial Analysis, with approximately 4.3 billion active parameters. According to the post, it is available on Hugging Face, and quantized versions for local use on RTX Spark are coming soon.

The same announcement mentions other updates for RTX Spark. A resource management control in the NVIDIA App lets users choose how GPU resources are shared between a foreground app and a background agent. NVIDIA also lists support for the LFM3 family of open models from Liquid AI.

Why it matters

NVIDIA notes that, because the weights are open, developers can choose where audio is processed, from a desktop to the cloud. This suggests that transcription could also run on a laptop, at least for people with suitable hardware.

NVIDIA says the model uses about 80% fewer active parameters than the No. 1 open weight model, and presents this as making high-quality transcription more practical on local hardware. These are the company's own claims, and we have not verified them independently.

The announcement is aimed first of all at developers, creators and gamers. It describes laptops built for those groups, not tools for small offices.

What it means for your business

For a small company, an agency or a professional, it is reasonable to expect that recording a client call or an internal meeting and getting a text with labeled speakers will become a more common option over the coming months. How soon, and at what quality in Italian, is something to check on your own recordings rather than assume from a press post.

A second point is where the audio ends up. Recordings of meetings and calls often contain personal data of clients or colleagues. The possibility of choosing where audio is processed is relevant, but it does not by itself settle questions about data protection. For those, it is sensible to ask your data protection adviser.

Third, a transcript is raw material, not a finished document. Someone still has to read it, correct names, numbers and dates, and turn it into a summary, minutes or a follow-up message. That work remains with the person, with the AI preparing the first draft.

What to do now

• Read the original NVIDIA post before making any decision on hardware, since the performance figures are the company's own.
• List the recordings your team handles and mark the ones that contain personal data, then ask a professional how they should be processed.
• If you already have a transcript of a meeting, paste it into the summarize text tool and compare the result with your own notes.
• Turn the agreed actions into a message to the client with the email reply tool, and check every name and date before sending.

At Infinithink we think the useful habit is the same whatever the source of the text, let the AI prepare the draft and decide yourself what goes out.

Sources

• NVIDIA, NVIDIA RTX Spark Windows Laptops Gain New Models, Agent Tools and Games

For information only, this is not legal or professional advice. Information is current as of the publication date and may change. Official sources checked on October 28, 2026.