Search AI Products and News

The ByteDance Doubao large model team announced the open-sourcing of UI-TARS-1.5, an open-source multimodal agent built on a vision-language model capable of efficiently executing various tasks in a virtual world. The model achieved state-of-the-art (SOTA) performance on seven typical GUI (Graphical User Interface) benchmark evaluations and demonstrated, for the first time, its long-term reasoning capabilities in games and interactive capabilities in open spaces. This open-source project marks a significant advancement in multimodal agent technology for GUIs.

Doubao recently released a series of updates to its large models. Doubao 1.5 Deep Thinking model and Doubao Text-to-Image 3.0 model are now officially available via Volcano Engine's open APIs for developers and enterprise clients. These two models have achieved industry-leading performance in both reasoning and general tasks, and have made significant progress in visual reasoning and image generation.

At the recently held Volcano Engine FORCE LINK AI Innovation Roadshow, Volcano Engine President Tan Dai officially announced the Doubao 1.5 Deep Thinking Model, marking the official launch of this innovative technology for enterprise services. Doubao 1.5 not only upgrades its basic functions but also introduces more intelligent application scenarios to help enterprises achieve more efficient digital transformation. Tan Dai stated at the launch that the Doubao 1.5 Deep Thinking Model can combine visual understanding to provide more innovative functions. For example, the model can generate...

Volcano Engine today released an event pre-heating announcement, officially declaring that its new Doubao large model will be officially launched tomorrow (April 17th) at the FORCE LINK AI Innovation Tour Hangzhou station. Official information shows that this release will bring a brand-new upgrade to the Doubao large model family and will particularly introduce the highly anticipated Doubao 1.5 Deep Thinking Model. Previous reports revealed that the Doubao APP and desktop client have undergone multiple rounds of experimental testing for the "Deep Thinking Mode." After this release, the model will officially provide services to enterprise clients. It is understood that Doubao...

Bug fixing is a persistent challenge in software development. ByteDance's Doubao large model team has exciting news: they've launched Multi-SWE-bench, the first multi-lingual software engineering (SWE) dataset. This new dataset aims to evaluate and improve the ability of large models to automatically fix code errors. Multi-SWE-bench significantly expands the scope compared to previous single-language datasets. It covers not only Python, but also...

SF Express Same City recently announced a comprehensive push towards digitalization and AI-driven decision-making across all operational aspects. The company aims to build a large model infrastructure tailored to the on-demand delivery industry for increased efficiency and improved service. Leveraging the DeepSeek open-source ecosystem and its multimodal AI capabilities, SF Express Same City can rapidly develop customized solutions. This allows for quick adaptation and adjustments to services and products based on specific client needs and market demands.

Doubao, the AI assistant from ByteDance, has announced the launch of its new deep thinking feature, now in testing. This update marks another breakthrough in AI assistants' information processing and reasoning capabilities, providing users with more comprehensive and accurate intelligent services.

Many self-media creators need to extract video transcripts, but good ones usually require payment, and the prices are not cheap. Some tools charge monthly subscriptions, ranging from tens to hundreds of yuan; others charge by the number of extractions, which adds up to a considerable expense. I discovered that the Doubao web version can actually extract transcripts for free. Simply enter the Douyin video link in the Doubao web version, click extract, and within seconds, the complete video transcript is ready, with high accuracy. For example, I uploaded a video previously posted by AIbase on...

This upgrade includes three major features: HTML preview, Python execution, and full project generation. First, Doubao now supports real-time preview and interaction with HTML code, allowing users to more intuitively create various mini-games and web pages on the platform, significantly improving development efficiency and user experience.

Amidst the booming development of the artificial intelligence field, Bytedance's Doubao large model team (Seed) recently held an all-hands meeting, marking a significant decision regarding the team's future direction. The meeting was co-hosted by Zhu Wenjia and the newly appointed head of AI fundamental research, Wu Yonghui. This was the first time the two leaders appeared together, attracting widespread attention from the industry. At the meeting, Zhu Wenjia and Wu Yonghui clearly stated that the Seed team's primary goal is to "explore the upper limit of intelligence," which will serve as the core guidance for the team's future work. They pointed out that exploring...

Today, the Doubao Large Model team officially released a technical report on its text-to-image technology, publicly disclosing for the first time the technical details of the Seedream2.0 image generation model. This encompasses the entire process, from data construction and pre-training framework to post-training RLHF, marking a significant breakthrough in the text-to-image field. Since its launch on the Doubao app and Jimeng in early December 2024, Seedream2.0 has served over 100 million C-end users and has become a favorite among professional designers. Compared to mainstream models like Ideogram2.0 and Midjourney V6.1, it...

ByteDance's Doubao large model team recently announced a breakthrough in addressing key bottlenecks in Mixture-of-Experts (MoE) architecture, open-sourcing a significant optimization technology called COMET. This technology dramatically improves large model training efficiency, achieving a remarkable 1.7x speedup and a 40% reduction in training costs. Image Note: Image generated by AI, image licensing provider Midjourney. COMET has been deployed in ByteDance's multi-thousand-GPU cluster training, resulting in millions of GPU hours saved.

Trae, China's first AI-native Integrated Development Environment (AI IDE), has officially launched domestically. Developed by ByteDance's technology team, Trae aims to provide developers with a more efficient and intelligent programming experience. The domestic version of Trae features Doubao-1.5-pro and supports switching between full-blooded DeepSeek R1 and V3 models, significantly improving programming speed and helping developers tackle complex technical challenges.