Apply for Trial
Newsroom

SenseTime Unveils SenseNova 5.5 - a Complete and Comprehensive Upgrade

2024-07-06

SenseTime, a strategic partner of the 2024 World Artificial Intelligence Conference & High-Level Meeting on Global AI Governance (WAIC 2024), held its AI Forum on "AI+: Catalyzing Next-Gen Transformations", where it unveiled the upgraded SenseNova 5.5 Large Model. The updates include SenseNova 5o, the first real-time multimodal model in China, which provides a new AI interaction model on par with GPT-4o’s streaming interaction capabilities.

 

SenseNova 5.5 has also launched an upgraded and cost-effective edge-side large model, reducing the cost per device to as low as RMB 9.90 per year, allowing for widespread deployment. Through continuous updates to its Cloud-to-Edge full-

stack large model product matrix, SenseTime provides innovative solutions for generative applications across various scenarios and industries. Currently, the SenseNova Large Model has been deployed at more than 3,000 government and corporate customers, including industries such as technology, healthcare, finance and programming.

 

XL press release.jpg

Dr. Xu Li, Chairman of the Board and CEO of SenseTime

 

Dr. Xu Li, Chairman of the Board and CEO of SenseTime, said: "This is a critical year for large models as they evolve from unimodal to multimodal. In line with users’ needs, SenseTime is also focused on boosting interactivity. With applications driving the development of models and their capabilities, coupled with technological advancements in multimodal streaming interactions, we will witness unprecedented transformations in human-AI interactions. "

 

Comprehensively Upgrading SenseNova 5.5: China’s First Real-time Multimodal Model

 

By gathering and processing data across modalities including audio, texts, images and videos, SenseNova 5o, China’s first real-time multimodal model provides a brand-new interactive AI experience.

 

Users are able to interact with SenseNova 5o, akin to having a conversation with an actual person. The interactive model is especially suitable for applications such as real-time conversation and speech recognition. It is highly adaptable and can manage multiple tasks within the same model, while adjusting its responses based on different contexts.

 

SenseNova 5.0, which was recently released in April, was China's first large model on par with the capabilities of GPT-4 Turbo. In quick succession two months later, the upgraded SenseNova 5.5 has registered a 30% improvement in overall performance compared to SenseNova 5.0. With significantly enhanced abilities in mathematical reasoning, English proficiency and following commands, SenseNova 5.5 interactivity and multiple core indicators are on par with GPT-4o.

 

SenseNova 5.5 adopts a hybrid cloud-edge collaborative expert architecture to maximize the “Cloud-to-Edge” synergy and reduce inference costs. The model training was based on over 10TB tokens of high-quality training data, including a large amount of synthetically-generated reasoning chain data, which help to enhance its reasoning capabilities.

 

To lower the barriers to entry for enterprise users in leveraging the robust capabilities of the SenseNova Large Model, SenseTime has recently launched the "Project $0 Go" scheme. This is a free and comprehensive onboarding bundle for all new enterprise users who are migrating from the OpenAI platform, including a 50 million tokens package and API migration consulting services.

 

A High-efficiency Cloud-based Model, with “Cloud-to-Edge” Full-stack Upgrades to Empower Enterprise Users

 

SenseTime has been active in promoting the research and development of edge-side large models, which can support the deployment of applications on multiple IoT devices such as smartphones, tablets, VR devices, in-vehicle computers, and smart lamps. By reducing the cost of each edge-side device to only RMB 9.90 per year, SenseTime expects to accelerate the adoption of the edge-side large model through cost-effectiveness, high availability and a low barrier to entry. At present, SenseTime has already initiated commercial partnerships with more than 150 customers.

 

SenseTime also further upgraded the edge-side large model and launched SenseChat Lite-5.5. This iteration features a significantly reduced inference time of 0.19 seconds, a 40% improvement over SenseChat Lite-5.0 that was earlier launched in April 2024. The inference speed has been increased by 15%, reaching 90.2 words per second, producing better results across the board.

 

In addition, SenseTime also launched an edge-side model product matrix, including specialized models such as the SenseChat Mini Writing Assistant, the Summary Assistant, and the Encyclopedia Assistant. With enhanced performance in their respective scenarios, customers can customize their preferred models to address their own business requirements.


New Additions to SenseNova’s Suite of Applications


As part of SenseNova 5.5, SenseTime has released Vimi, its first controllable AI avatar video generator. With just a single photo, Vimi can generate short video clips with precise control over an avatar's facial expressions and upper body movements. It realistically renders changes in lighting, shadows, hair, clothing, and backgrounds, ensuring a cohesive final result. Vimi can generate videos up to one minute long without any loss in quality, making it a reliable tool for long-form video generation in entertainment and interactive applications.


Building on the “Cloud-to-Edge” full-stack deployment, SenseTime has been constantly upgrading and expanding generative AI applications under the SenseNova Large Model Series to meet the needs of more users and empower industries in their digital transformations.

 

The SenseTime Raccoon Series, the SenseNova AI Native productivity tool, has also been upgraded. The Code Raccoon (Consumer Edition Upgraded) has been released, with a five-fold improvement in response speed and a 10% increase in coding precision. It also has enhanced model capabilities, a richer set of plug-ins and a more comprehensive data dashboard. Office Raccoon now has a consumer-facing webpage and a WeChat mini-app version, allowing users to directly upload, analyze and process files in WeChat for greater efficiency and convenience.


SenseTime has continued to capitalize on AI’s new capabilities to empower vertical industries such as finance, agriculture, cultural tourism, and healthcare, embedding AI in their business operations to boost productivity and cost-efficiency:


l  Large model for the financial sector: The finance-based AI system Agent is equipped with professional vertical agent capabilities to improve quality and efficiency in areas such as compliance, marketing, data development, and investment research;

l  Large model for the agricultural sector: Helps increase efficiency of agricultural analysis by more than five times, reduces use of agricultural materials by 20%, and increases crop yields by 15%;

l  Large model for cultural tourism: Enhances travel planning efficiency by more than eight times, ticket bookings by 4.5 times faster, and data analysis a thousand times more efficiently


2024 is seen as a pivotal year for the applications of large models, which coincides with SenseTime’s 10th anniversary. SenseTime’s journey over the past decade has culminated in the development of a comprehensive full-stack large model product matrix covering Cloud-to-Edge. Looking ahead, SenseTime will continue to expand the SenseNova industry ecosystem  to empower even more businesses and communities in their digital transformation journeys.