
DoubaoLarge ModelIt is a family of models with multimodal capabilities introduced by ByteDance, covering several models with different technical features and highlights. Originally named "Lark", ByteDance's Big Model was officially released on May 15, 2024 at the Volcano Engine Power Conference. The model is one of the first big models to pass the Interim Measures for the Administration of Generative Artificial Intelligence Services, meaning that its technology and application meet the requirements of the relevant regulations.
1. Model family members
The Beanbag Big Model family mainly consists of the following members:
- Beanbag General Model Pro: for complex applications requiring deep text understanding and generation.
- Beanbag Generic Model Lite: more cost-efficient and suitable for scenarios with stringent requirements on speed and running costs.
- Beanbag-Roleplay Modeling: the ability to simulate different roles in a conversation.
- Beanbag-Speech Synthesis Model: provides natural speech synthesis technology.
- Beanbag-voice replica modeling: highly reproducible voice replication technology.
- Beanbag-Speech Recognition Model: for converting speech to text.
- Beanbag-text-generated graph model: the ability to generate images that match the textual content.
- Beanbag-Function Call model: specific functions and application scenarios may involve more specialized technical calls.
2. Technical characteristics
- multimodal capability: The beanbag big model family is not limited to processing text, but also covers multiple modalities such as language, vision and sound, enabling cross-modal information understanding and interaction.
- Customization & Personalization: The model design takes into account the needs of different industries and business scenarios, and supports a high degree of customization and personalization.
- High performance and low latency: Demonstrates low latency and high throughput when processing large-scale data, ensuring performance in real-world applications.
- Safety and reliability: Multi-dimensional security measures are taken to ensure the safe and stable operation of the model.
3. Application scenarios
The Beanbag Big Model family has been applied in multiple business scenarios both internally and externally, significantly improving efficiency and product experience. These scenarios include, but are not limited to, more than 50 businesses such as Jitterbug, Tomato Novels, Flying Book, and Mega Engine.
4. Data-processing capacity
The Beanbag Big Model processes 120 billion Tokens of text and generates 30 million images on a daily basis, and is becoming one of the most heavily used big models with the richest application scenarios in China.
With its multi-modal capabilities, customization and personalization, high performance and low latency, secure and reliable technical features, as well as a wide range of application scenarios and competitive pricing strategies, Beanbag Big Model is becoming one of the most talked about Big Models in the industry.
data statistics
Relevant Navigation

OpenAI's large-scale language model, officially launched on February 28, 2025, is an upgraded version of GPT-4.

Kling LM
Racer's self-developed advanced video generation model supports the generation of high-quality videos based on text descriptions, helping users to efficiently create artistic video content.

TranslateGemma
Google's open source lightweight multimodal translation model supports 55 languages and image translations, with performance that exceeds larger models, taking into account both mobile and cloud deployments, and facilitating efficient globalized communication.

LingBot-Depth 2.0
Ant Group has launched a new-generation spatial perception model for robots that enhances 3D depth perception capabilities and improves performance in grasping, navigation, and environmental understanding.

Congrong LM
The multimodal large model independently developed by CloudScience has the ability of real-time learning, synchronous feedback, cross-modal interaction, etc. It is widely used in many industries such as finance, security, government affairs, etc., to promote the popularization and development of AI applications.

Chitu
The Tsinghua University team and Qingcheng Jizhi jointly launched an open source large model inference engine, aiming to realize efficient model inference across chip architectures through underlying technological innovations and promote the widespread application of AI technology.

Seedream 2.0
Byte Jump launched a native bilingual image generation model with excellent comprehension and rendering capabilities for a wide range of creative design scenarios.

Qwen3-Next
Ali open source 80 billion parameters of the big model, 1:50 super sparse activation, millions of contexts, the cost down 90%, the performance is comparable to the hundreds of billions of models.
No comments...
