GitHub radar

Open agentic model for browser and terminal control

Nex-AGI released an open-weights 35B model that autonomously browses the web, runs code, and completes multi-step agentic tasks.

01nex-agi/Nex-N2.5-mini 7935.2k downloads/mo35B paramsimage-text-to-text

Nex-AGI built a 35B open-weights multimodal model for agentic tasks. The authors describe it as 'a next-generation family of agentic models built for long-horizon tasks in real-world environments.' It can browse the web, operate a terminal, execute and test code, understand images, and call external tools. The model supports three reasoning depths — fast, medium, and deep — depending on task complexity. On OSWorld-Verified it scores 71.2 and on SWE-Bench Pro 43.8. Quantized versions are available for llama.cpp, Ollama, and LM Studio.

Why a vibe-coder should care

If you want to experiment with an open computer-use agent, this is one of the most capable open-weights options available now. Be honest about the hardware requirement: the quantized version weighs around 21 GB, so you need a Mac with 64 GB of unified memory or a rented cloud GPU — a typical laptop won't manage it.

How to install

Copy this and send it to your agent — Claude Code, Codex, any of them:

Find a quantized version of Nex-AGI's Nex-N2.5-mini that fits my RAM and install it via Ollama or LM Studio: https://huggingface.co/nex-agi/Nex-N2.5-mini — run it and show me how to use it

The 4-bit quantized version requires around 21 GB — a Mac with 64 GB unified memory or a rented cloud GPU. The authors officially recommend a dual H100 setup.

Open on Hugging Face