Skip to content
NewA calmer, faster ZeroHour

Nex-AGI releases Nex-N2.6, a 321B agent model

Original title: nex-agi/Nex-N2.6 — new model trending #23 on Hugging Face

Hugging Face trending modelsby nex-agi ()2 min read
AI summary · x-ai/grok-4.7Summaries by Vani AI

The FP8 weights, built on GLM-5.3-Flash, score 83.0 on OSWorld in team-reported tests.

Nex-AGI published FP8 weights for Nex-N2.6, a multimodal agent model of about 321B parameters built on GLM-5.3-Flash. The release targets computer use, coding and long-horizon tool use, and is offered through OpenRouter. The team reports less repetitive reasoning than Nex-N2.5-Pro and an OSWorld score of 83.0.

Unconfirmed: Benchmark scores are reported by Nex-AGI and depend on harness settings.

Key facts
  • Reasoning repetition fell from 16.0 percent to 1.6 percent internally.
  • Tool-list adherence was 100 percent when responses made tool calls.
  • Reported OSWorld score is 83.0 versus 82.2 for Nex-N2.5-Pro.
  • Weights are FP8 and intended for an FP8-compatible SGLang setup.
Explain this AI news with Vani AIWhat was announced, what is new and why it matters, in a new chat

From the original article

2 min read at huggingface.co

image-text-to-text · 916 downloads · 139 likes · license apache-2. 0

Hugging Face trending models

By nex-agi · huggingface.co

Read the full article at Hugging Face trending models

A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.