ByteDance DouBao Launches Seed 2.1 Series: Three Indicators of Coding and Agent Capabilities Comparable to GPT-5.5

ByteDance has quietly launched the DouBao Seed 2.1 series models on the Volcano Engine official website, including the Pro and Turbo versions, while also updating the Seed-Evolving model. These new models are officially positioned as "the next-generation large model designed for the Coding and Agent era," targeting two core scenarios: complex engineering delivery and large-scale production.

Compared to the previous generation, Seed 2.1 has made comprehensive upgrades in three directions: Coding engineering delivery, Agent long-chain task execution, and multimodal understanding. With stronger autonomous planning and dynamic repair capabilities, it is capable of handling real R&D and high-value production tasks. The official claims that these three core capabilities are comparable to GPT-5.5.

Pro flagship and Turbo efficient, each with its own role

The positioning of the two models is clear. Doubao-seed-2-1-pro is a flagship deep thinking model, focusing on strong demand understanding, long-term planning, and continuous repair capabilities, suitable for high-complexity exploration scenarios such as complex coding, long-chain agents, and multi-step engineering delivery; Doubao-seed-2-1-turbo is a low-cost and low-latency version for large-scale production, with complete features and performance comparable to the Pro version, specifically designed for enterprise-level deployment that needs to handle a large number of online calls.

In addition, Doubao-seed-evolving uses a dynamic iteration mechanism, releasing at least one version update per week. Users can always call the latest version through a unified Model ID, ensuring continuous evolution without interruption. There is also a role model, Doubao-seed-character, which has been fully upgraded in general entertainment scenarios, adding capabilities such as natural conversation, video understanding, and multimodal sticker recognition, and supports active activation of conversations.

Regarding pricing, the Pro version's input is 6 yuan per million tokens, and output is 30 yuan per million tokens; the Turbo version's input is 3 yuan per million tokens, and output is 15 yuan per million tokens. Compared to the Pro version, the Turbo version's price is halved, providing a more competitive cost structure for large-scale calls.

Yingmou Technology Secures Hundreds of Millions in New Round of Funding, Unveils Rodin Gen-2.5, a Million-Face-Level 3D Large Model

Yingmou Technology has completed a funding round worth hundreds of millions of yuan, led by Kaifu Fund and Shanghai GuoTou Daxiang. The funds will be used for 3D large model research and development and global commercialization, accelerating the implementation of game and e-commerce scenarios. The core product Hyper3D has been upgraded, with 80% of revenue coming from overseas, serving clients such as ByteDance and Unity.

Volc Engine Launches Doubao Seedance 2.5 Video Model, the Azuris Large Model Service Has Over 1.1 Million Users

On June 23, 2026, Volcano Engine launched its Seedance2.5 video generation model at the Summer FORCE Conference, set to release in July. It offers three key breakthroughs: direct generation of 30-second native videos, joint generation from up to 50 multimodal assets, and local editing with visual consistency. President Tan Dai noted that video generation is crucial to world models.....

ByteDance Volcano Engine 2026 Conference Launches: Seedance 2.5 Directly Outputs 30-Second Videos, Doubao 2.1 Pro Competes with Opus 4.6

ByteDance launches the video generation model Seedance 2.5, which supports directly outputting a complete 30-second video in one go, marking the entry of video generation into the long-sequence era. At the same time, it also introduces the multimodal model Doubao 2.1 and the image model Seeddream 5.0, enhancing its competitiveness in the AI field.

New Powerhouse for Film and Television Production: Shengshu Vidu Q3 Launched on Huawei Cloud, Creating a Video Generation Solution Designed for Dramas

Huawei Cloud partners with Shengshu Technology to launch the world's first video large model Vidu Q3 designed specifically for dramas. It focuses on integrated text-to-video and image-to-video generation. The model is tailored for enterprise marketing, film and television, and cultural and creative design. Its core advantage lies in strong narrative capabilities, enabling the efficient creation of coherent video content and accelerating the realization of creative ideas.

The Fifth Listing Criterion of the Sci-Tech Innovation Board Opens to Large Model Enterprises, with the Capital Market Accelerating the Development of AI Innovation

On June 17, 2026, the Shanghai Stock Exchange issued guidelines to implement the CSRC's 2025 opinions on the STAR Market growth tier. By expanding the fifth set of listing criteria, it prioritizes supporting the listing of general and industry large model firms with no substantial revenue yet but outstanding potential.....