88Z AI, 29 April — Ant Group's Bailin has officially open-sourced Ling-2.6-flash, offering multiple precision versions on major platforms. The model has 104B total parameters and previously garnered community attention under an anonymous identity.
At a glance
- Ling-2.6-flash is officially open-sourced with BF16, FP8, and INT4 precision versions
- The model has 104B total parameters and previously gained attention as Elephant Alpha
The update
Ant Group's Bailin large model has announced the official open-source release of its efficient Instruct model, Ling-2.6-flash. To accommodate diverse deployment needs, multiple precision versions including BF16, FP8, and INT4 are provided simultaneously. The model weights are now available for download on Hugging Face and ModelScope. The model features 104 billion total parameters and 7.4 billion activated parameters. Notably, two weeks prior, it appeared on 第三方模型接口 under the anonymous identity Elephant Alpha, quickly attracting significant attention and discussion within the developer community and among industry users due to its performance. This official open-source release marks its transition from anonymous testing to public application.
Sources & editorial note
This article is an AI-assisted synthesis by 88Z AI based on the public sources below. Product claims and availability are attributed to their publishers. It does not represent independent testing or first-hand reporting.
- 百灵大模型Source published
Source dates above refer to the original announcements; the date under the headline is this article’s publication on 88Z AI.