Multimodal model

Nex N2.5 mini

A 35 billion parameter Mixture of Experts multimodal agentic model that operates computers and browsers and autonomously executes and tests programs, using vision as feedback. Supports a 262144 token context window and runs on two H100 GPUs.

Type
Multimodal
Weights
Open
Downloads
7.3k
Likes
828
Pipeline
text generation
License
apache-2.0

Model hub

nex-agi/Nex-N2.5-mini

Repo
nex-agi/Nex-N2.5-mini
Library
transformers
Access
Public
Technical tags and languages
transformers safetensors qwen3_5_moe image-text-to-text text-generation conversational endpoints_compatible
Open on Hugging Face

More from Nex-AGI

4

About Nex-AGI

1

Nex-AGI develops open-weight large language models built for agentic tasks, including its Nex-N2.5 model family (Mini, Pro, Max) that combine vision and action for computer use, web browsing, and software engineering agents.