Holotron-12B - High Throughput Computer Use Agent

Holotron-12B - High Throughput Computer Use Agent

Hugging Face Blog 5 min read

About this article

A Blog post by H company on Hugging Face

Back to Articles Holotron-12B - High Throughput Computer Use Agent Team Article Published March 17, 2026 Upvote 12 +6 Pierre-Louis Cedoz plcedoz38 Follow Hcompany Hamza Benchekroun hamza-hcompany Follow Hcompany Aurélien Lac h-aurelien-lac Follow Hcompany delfosse aureliendelfosseathai Follow Hcompany Tony Wu h-tonywu Follow Hcompany Mats L. Richter MatsLRichter Follow Hcompany Antoine Bonnet ABonnetH Follow Hcompany Kai Yuan h-kaiy Follow Hcompany Aleix Cambray (H-AI) h-aleixcambray Follow Hcompany Alexandra a-constantinou Follow Hcompany We're thrilled to release Holotron-12B, a multimodal computer-use model from H Company. Post-trained from the open NVIDIA Nemotron-Nano-2 VL model on H Company’s proprietary data mixture, Holotron-12B is the result of a close collaboration between our research labs to engineer a new type of model optimized primarily for scale and performance in production. H Company is part of the NVIDIA Inception Program. The model is now available on Hugging Face. Why We Built Holotron-12B Most multimodal models today optimize primarily for static vision or following instructions. Holotron-12B, just like our Holo2 model, however, has a different goal: serving as a policy model for computer-use agents that must perceive, decide, and act efficiently in interactive environments. With Holotron-12B, we wanted to create a model that could efficiently and effectively scale in production while handling long contexts with multiple images, and still perform well...

Originally published on March 17, 2026. Curated by AI News.

Related Articles

[2603.25112] Do LLMs Know What They Know? Measuring Metacognitive Efficiency with Signal Detection Theory
Llms

[2603.25112] Do LLMs Know What They Know? Measuring Metacognitive Efficiency with Signal Detection Theory

Abstract page for arXiv paper 2603.25112: Do LLMs Know What They Know? Measuring Metacognitive Efficiency with Signal Detection Theory

arXiv - AI · 4 min ·
[2603.24772] Evaluating Fine-Tuned LLM Model For Medical Transcription With Small Low-Resource Languages Validated Dataset
Llms

[2603.24772] Evaluating Fine-Tuned LLM Model For Medical Transcription With Small Low-Resource Languages Validated Dataset

Abstract page for arXiv paper 2603.24772: Evaluating Fine-Tuned LLM Model For Medical Transcription With Small Low-Resource Languages Val...

arXiv - Machine Learning · 4 min ·
[2603.25325] How Pruning Reshapes Features: Sparse Autoencoder Analysis of Weight-Pruned Language Models
Llms

[2603.25325] How Pruning Reshapes Features: Sparse Autoencoder Analysis of Weight-Pruned Language Models

Abstract page for arXiv paper 2603.25325: How Pruning Reshapes Features: Sparse Autoencoder Analysis of Weight-Pruned Language Models

arXiv - AI · 4 min ·
Llms

[D] Why evaluating only final outputs is misleading for local LLM agents

Been running local agents with Ollama + LangChain lately and noticed something kind of uncomfortable — you can get a completely correct f...

Reddit - Machine Learning · 1 min ·
More in Open Source Ai: This Week Guide Trending

No comments

No comments yet. Be the first to comment!

Stay updated with AI News

Get the latest news, tools, and insights delivered to your inbox.

Daily or weekly digest • Unsubscribe anytime