Ant Group Releases Ling-2.5-1T and Ring-2.5-1T Open-Source AI Models

Feb 16, 2026
Ant Group has announced Ling-2.5-1T, a trillion-parameter large language model, and Ring-2.5-1T, a hybrid linear-architecture reasoning model, expanding its open-source Ling model family. Both models are available on Hugging Face and ModelScope.

Ant Group has released two new open-source AI models, Ling-2.5-1T and Ring-2.5-1T, expanding its Ling model family. The models are available under open licenses on Hugging Face and ModelScope.

Ling-2.5-1T is a trillion-parameter large language model designed for higher reasoning efficiency, fine-grained preference alignment, and native agent interaction. It supports context lengths of up to one million tokens and achieves comparable performance to leading models while using fewer tokens.

Ring-2.5-1T, described as the world’s first hybrid linear-architecture reasoning model, is optimized for advanced reasoning tasks. It has achieved gold-tier scores on academic benchmarks, including 35 out of 42 on the International Mathematical Olympiad 2025 and 105 out of 126 on the Chinese Mathematical Olympiad 2025. The model reduces memory access by more than ten times and increases throughput by over three times compared to its predecessor in long-text generation scenarios.

Both models are part of Ant Group’s broader open-source Ling (BaiLing) ecosystem, which includes non-thinking, reasoning, and multimodal series. The company also recently introduced Ming-Flash-Omni-2.0, a unified speech, audio, and music model. Together, these releases mark a continued expansion of Ant Group’s open AI model offerings.

We hope you enjoyed this article

Consider subscribing to one of our newsletters like Daily AI Brief.

Also, consider following us on social media:

Free newsletter

Daily AI Brief

Daily report covering major AI developments and industry news, with both top stories and complete market updates

Whitepaper

Tensordyne Napier: What If One Rack Could Do the Work of Nine?

Tensordyne

This Tensordyne whitepaper presents Napier, an inference-focused AI processor and rack-scale system based on the company’s TDN Math logarithmic number system. It examines infrastructure requirements for large mixture-of-experts and agentic models, compares major inference architecture approaches, and details the TDN AIP processor, TDN72 pod, TDN Link fabric, and Napier Ultra configuration. The paper reports simulation-based performance, cost, and accuracy-validation results, including Tensordyne’s projected comparison of one Napier rack with a nine-rack Nvidia Rubin plus Groq deployment; the chip is reported as taped out and in fabrication.

Read more
Free, six days a week

Daily AI Brief: the AI news that matters, in your inbox.