# ModelScope adds Ling-3.1-flash-base with 1M context and 560B parameters

> Original title: Ling-3.1-flash-base is now on ModelScope with 1M context, 560B total parameters, and 25B active per token.

- Source: ModelScope
- Published: 2026-10-27T12:38:31.000Z
- Bad Influence: https://badinfluence.news/items/q1e7dgxdhrnhic5kfj2doxecq
- Original: https://x.com/ModelScope2022/status/2115060574950244553

## Summary

ModelScope says Ling-3.1-flash-base is now available on its platform. The base model has a 1M-token context window, 560B total parameters, and about 25B active parameters per token. The post says it is meant for continued pre-training, supervised fine-tuning, and domain-specific post-training under an MIT License.

## Post

Ling-3.1-flash-base is now on ModelScope with 1M context, 560B total parameters, and 25B active per token.
🤖 https://modelscope.ai/models/inclusionAI/Ling-3.1-flash-base

🧠 Built to be adapted: Start from the base checkpoint for continued pre-training, supervised fine-tuning, and domain-specific post-training.
📚 Built for long-horizon work: A 1M-token context window gives custom systems room to work across long documents, codebases, and multi-step workflows.
⚙️ Large capacity, sparse compute: Its hybrid MoE architecture stores 560B parameters while activating about 25B parameters per token.
🛠️ Make it your own: Shape domain terminology, output conventions, task behavior, and assistant policies for the workflows you need.  
📜 MIT License.

![](https://badinfluence.news/api/img-proxy?u=https%3A%2F%2Fpbs.twimg.com%2Fmedia%2FHVo0Yj9WgAAEPWR.jpg&mode=thumb&exp=1793404800&sig=c1f1f4c401bab624)