摩尔线程官宣:MTT S5000 适配字节跳动 Protenix-v2 开源生物分子结构预测模型Moore Threads officially announces: MTT S5000 adapted for ByteDance's Protenix-v2 open-source biomolecular structure prediction model


AI free token/AI免费积分
摩尔线程官宣:MTT S5000 适配字节跳动 Protenix-v2 开源生物分子结构预测模型
感谢IT之家网友 Autumn_Dream 的线索投递!

IT之家 9 月 24 日消息,摩尔线程今天宣布,旗下训推一体智算卡 MTT S5000 正式完成了开源生物分子结构预测模型 Protenix-v2 的全链路推理兼容。

据IT之家了解,Protenix-v2 开源模型由字节跳动 Seed 团队研发,发布于 2026 年 4 月,在多项核心基准测试中表现持平甚至超越 DeepMind AlphaFold 3,是当下全球少数可以对蛋白质、DNA、RNA 及小分子配体复合物开展高精度端到端预测的开源基座。

MTT S5000 此次适配,标志着摩尔线程达成了大分子结构预测、全基因组序列建模与临床医学影像三大 AI for Science(AI4S)核心情形的端到端工程验证,为国内药企及科研机构给出 AI4S 算力底座。

本文来源:https://www.ithome.com/1/006/989.htm
若有版权问题,请来信告知:13951431913#139.com 或在右方提交评论。

Thanks to IT之家 netizen Autumn_Dream for the tip!

IT之家, September 24 - Moore Threads announced today that its training-inference integrated AI computing card MTT S5000 has officially completed full-chain inference compatibility with the open-source biomolecular structure prediction model Protenix-v2.

According to IT之家, the Protenix-v2 open-source model was developed by ByteDance's Seed team and released in April 2026. It performed on par with or even surpassed DeepMind AlphaFold 3 in multiple core benchmark tests, and is currently one of the few open-source foundations in the world capable of high-precision end-to-end prediction of protein, DNA, RNA, and small-molecule ligand complexes.

This adaptation of the MTT S5000 marks Moore Threads' completion of end-to-end engineering validation across three core AI for Science (AI4S) scenarios - macromolecular structure prediction, whole-genome sequence modeling, and clinical medical imaging - providing an AI4S computing power foundation for domestic pharmaceutical companies and research institutions.

Moore Threads officially announces: MTT S5000 adapted for ByteDance's Protenix-v2 open-source biomolecular structure prediction model

暂无评论,快来发表第一条评论吧!No comments yet. Be the first!

📮 需求咨询📮 Inquiry