Nex-AGI Releases Nex-N2.5-mini 35B Open-Weights Model
Nex-AGI has announced the release of Nex-N2.5-mini, a 35-billion parameter open-weights multimodal language model. It is part of the new Nex-N2.5 agentic model family, alongside larger Pro and Max variants. The model brings focused enhancements in computer use, web browsing, and visually grounded capabilities to open-source AI deployments. This allows developers to run capable agentic AI systems locally without depending on proprietary cloud APIs. Nex-N2.5-mini leverages an adaptive thinking mechanism to optimize chain-of-thought reasoning, saving approximately 20% in token expenditure while matching or exceeding standard chain-of-thought performance. It expands upon the multimodal foundations introduced in the earlier Nex-N2 model series.
## BACKGROUND
Agentic AI models are designed to interact with software, browse websites, and execute complex task workflows autonomously beyond traditional text generation. Open-weights models allow researchers and developers to download, modify, and host AI architectures locally on their own infrastructure.