ByteDance Releases 'Doubao'  LLM: 0.7% of GPT4 Price at on Par Performance

ByteDance Releases 'Doubao' LLM: 0.7% of GPT4 Price at on Par Performance

By
Yumi Takahashi
2 min read

ByteDance Unveils Doubao LLM Through Lingyun Engine, 0.7% of the GPT4 Price at On Par Performance

ByteDance, the parent company of TikTok, has unveiled its self-developed LLM, "Doubao," through its cloud service platform, Lingyun Engine. Originally named "Yunque," Doubao is among the first domestic models to be algorithmically registered in China. It has already been used in ByteDance's AI chat assistant and will now be available through Lingyun Engine. This announcement follows similar moves by Alibaba Cloud and Tencent Cloud, albeit later by a year and a half. The expansion of Lingyun Engine's cloud services in 2023 marks a significant shift, including the release of its self-developed DPU series cloud products, upgraded machine learning platforms, and large model training support.

ByteDance claimed that users can purchase 1.25 millions tokens for one RMB ($0.13), which is only 0.7% of the price of GPT4.

Key Takeaways

  • ByteDance's cloud service platform, Fire Mountain Engine, has debuted their self-developed "Doubao" large model, now named "Doupackage."
  • Doupackage is one of the first domestic models to be algorithmically registered in China and is already integrated into ByteDance's AI chat assistant "Doupai."
  • Fire Mountain Engine will now offer Doupackage as a service, following in the footsteps of competitors such as Alibaba Cloud and Tencent Cloud.
  • Fire Mountain Engine expanded their cloud services from the model production end to the model use end in 2023.
  • They released a series of cloud products, including their own DPU and an upgraded machine learning platform that supports training of large models.
  • Doubao, together with other Chinese LLMs of high performance and low price like DeepSeek V2, signals a new US-China war on LLM.

Analysis

ByteDance's introduction of its Doubao model through Lingyun Engine marks its entry into the AI cloud market, aligning with moves made by industry giants like Alibaba Cloud and Tencent Cloud. This strategic expansion enhances ByteDance's AI capabilities, fueling innovation within the company's AI chat assistant and related services. The potential outcomes include heightened competition, augmented R&D investments, and potential collaborations with enterprises seeking AI solutions. Long-term implications encompass a more diverse AI landscape in China, propelling technological advancements and economic growth. However, this development may pose challenges for smaller market players due to increased competition for industry consolidation and talent acquisition.

It also signals a newly created US-China battlefield: LLM. We are very certain that the extra cheap and performant Chinese LLM on par with GPT4 performance will create a huge shock to their American counterparts.

Did You Know?

  • Doubao/Doupackage: A large model developed by ByteDance's cloud service platform, Fire Mountain Engine. It is among the first domestic models to be algorithmically registered in China and has been utilized in ByteDance's AI chat assistant "Doupai." Fire Mountain Engine will now provide Doubao (Doupackage) as a service, following the footsteps of competitors like Alibaba Cloud and Tencent Cloud.
  • Lingyun Engine/Fire Mountain Engine: ByteDance's cloud service platform, which expanded its cloud services from model production to model utilization in 2023. It released self-developed DPU series cloud products, an enhanced machine learning platform, and began supporting large model training. In addition, it introduced "Huoshan Fangzhou" in June, offering model services by incorporating other companies' models.
  • DPU (Deep Learning Processing Unit): A type of cloud product released by Lingyun Engine, specialized hardware accelerators designed to expedite the processing of deep learning models. These models are integral to various AI applications. The release of DPU series cloud products aims to enhance the performance and efficiency of Lingyun Engine's cloud services for AI and machine learning workloads.

You May Also Like

This article is submitted by our user under the News Submission Rules and Guidelines. The cover photo is computer generated art for illustrative purposes only; not indicative of factual content. If you believe this article infringes upon copyright rights, please do not hesitate to report it by sending an email to us. Your vigilance and cooperation are invaluable in helping us maintain a respectful and legally compliant community.

Subscribe to our Newsletter

Get the latest in enterprise business and tech with exclusive peeks at our new offerings