Open agentic model family from Nex AGI for computer use, coding and long-horizon tool tasks, in several sizes.
Updated EU status checked Last shown
Sources: the vendor's own pages (linked above) and creator videos at the linked timestamps. EU status is information, not legal advice; see how we check.
Nex N2 is a model family from Nex AGI aimed at agentic work: operating computers and browsers, coding, and long-horizon tool use. According to a YouTube creator summary, the N2 generation was trained from Qwen 3.5 with adaptive reasoning, with a Pro variant of 397B parameters (17B active) and a Mini variant of 35B (3B active).
The official site now presents later updates. Nex-N2.5 comes in Mini, Pro and Max sizes. Mini and Pro keep the multimodal base of N2 and focus on computer use, web browsing and visually assisted agency. Max is built on DeepSeek-V4-Pro-Base, a 1.6T-parameter text-only MoE. Nex-N2.6 is described as the latest agentic model, built on GLM-5.3-Flash with about 321B parameters, and it is said to reduce repetitive reasoning and follow the tools available in the current environment more closely than Nex-N2.5-Pro.
The site lists the models under "Open-Source Models" and describes NexRT, a single-request inference engine for Nex-N2.5-Pro on 8 H100 or H200 GPUs, supporting standard decoding, MTP and DFlash, with SGLang handling prefill. Coding evaluations use the NexAU harness, and computer-use evaluations use NexCUA, which the vendor says will be open-sourced. Weight download links, licence terms and pricing were not included in the provided material, and benchmark figures should be checked against the vendor's own reports.
Watch the moment: the exact second top creators show it. Videos load only when you press play, via youtube-nocookie.com.
Open model family trained from Qwen 3.5 with adaptive reasoning, beating DeepSeek-V4 and GLM-5.1 on agentic and coding benchmarks. Pro is 397B (17B active), Mini is 35B (3B active).
Closest neighbours by meaning, not just by category.
App
Multi-model AI workspace giving access to ChatGPT, Claude, Gemini and other models, with a no-code agent builder.
Model · Nvidia
NVIDIA's family of open models for agentic AI: reasoning, vision, speech, retrieval and safety models.
Model · BAAI
27B open-weights agent model from BAAI that iteratively proposes, tests and refines solutions on coding and ML tasks.
Model · NVIDIA
NVIDIA's open-weight 30B mixture-of-experts model with 1M-token context, aimed at long-running agent workloads.
Model
35B Mixture-of-Experts open-weights model from InternScience for long-horizon agentic search, engineering and tool use.
Model · Meituan
Meituan's open-weights 1.6T-parameter multimodal MoE model (48B active) for agentic coding and visual tasks.
