The Doubao 2.1 Pro model has been released, with Coding and Agent capabilities crossing a production-grade tipping point, and multiple benchmarks ranking in the international first tier.
On June 23, Volcano Engine held its 2026 Summer FORCE Power Source Conference in Beijing, where it released Doubao Large Language Model 2.1, along with several new models for video, image, and audio, and upgraded its cloud service system designed for Agents.
According to Volcano Engine, as of June this year, the daily average token call volume for the Doubao Large Language Model has surpassed 180 trillion, growing more than 10-fold over the past year. IDC data shows that in China's public cloud MaaS service market, Volcano Engine ranks first with a market share of 49.5%.
Doubao 2.1 Pro Crosses the Production-Grade "Inflection Point"
Volcano Engine President Tan Dai stated that only when model capabilities cross the "inflection point" can they truly meet the needs of enterprises and individuals in production scenarios. Globally, the first video generation model to cross the inflection point is Seedance 2.0, while in the Coding and Agent fields, it is Claude Opus 4.6.
As the flagship model released at this conference, Doubao 2.1 Pro has achieved capability leaps in three core areas: Coding, Agent, and VLM (Vision-Language Model), with multiple benchmark results outperforming Claude Opus 4.6, officially crossing the production-grade inflection point.
Benchmark results show that Doubao 2.1 Pro ranks in the top tier in coding evaluations such as Terminal Bench 2.1, SWE-Pro, and SciCode; in Agent and multimodal aspects, the model also ranks among the global leaders in evaluations including OSWorld, MobileWorld, and MMMU-Pro.
3 images · Swipe left/right to browse
Tan Dai demonstrated the model's Coding and Agent capabilities with practical cases: in a chip design RTL test, Doubao 2.1 Pro ran continuously for nearly 18 hours, underwent 9 rounds of iteration, and successfully completed full engineering workflows including simulation, testing, and synthesis checks, showcasing the model's production-grade Coding delivery capability in real engineering scenarios; using the Doubao 2.1 Pro model to build a 3D virtual city scene, it enabled over 500 intelligent Agents to collaborate simultaneously, complete over a thousand rounds of tool calls, and generate over a hundred buildings.
It is reported that Doubao 2.1 Pro is priced at 6 yuan per million tokens for input, 30 yuan for output, and only 1.2 yuan for cache hits. Volcano Engine stated that its comprehensive usage cost is nearly 80% lower than Claude Opus 4.6. For high-frequency call scenarios, the Doubao 2.1 Turbo is priced at half the cost of 2.1 Pro, offering even better cost-effectiveness. Additionally, Volcano Engine has launched the Doubao-Seed-Evolving version, targeting Coding and Agent scenarios, with rapid iteration expected at a frequency of 2 to 4 times per month.
Currently, the Doubao 2.1 model has opened API services on Volcano Engine and is being progressively integrated into products such as Doubao, TRAE, and Kouzi.
Seedance 2.5 Debuts, Expanding Applications in Physical Industries
At this FORCE conference, the latest Doubao video generation model, Seedance 2.5, made its debut and is expected to be officially launched in July.
According to the introduction, Seedance 2.5 has achieved three capability improvements: native generation of 30-second single-segment videos in one pass, support for joint generation from up to 50 full-modality assets, and local editing that maintains visual consistency.
Tan Dai believes that video generation is one of the pathways to world models and holds enormous application potential in physical industries. Currently, Seedance has been deployed in areas such as embodied intelligence, industrial manufacturing, and intelligent driving, providing new tool capabilities for business needs including data synthesis, scenario simulation, and process demonstration.
In addition, Volcano Engine also showcased the Doubao image generation model Seedream 5.0 Pro and the Doubao audio generation model Seed-Audio 1.0, further consolidating its leadership in multimodal capabilities.
According to the introduction, Seedream 5.0 Pro features capabilities such as interactive precise editing, multi-layer separation, high-density information expression, and native multilingual text generation, and can directly generate editable layered design files and complex infographics. Seed-Audio 1.0 features zero-shot multimodal reference capability, supporting one-shot generation of multi-character dialogue, background music, and sound effects.
For the Agent era, Volcano Engine has once again upgraded its AI cloud-native architecture, releasing the Ark CLI command-line tool, which allows developers to connect Agents to Volcano Ark with a single line of command; it also upgraded AgentKit and HiAgent 3.0, and launched the ArkClaw enterprise Agent workbench and the AI Trust product suite, enhancing the infrastructure and security capabilities for enterprise Agent applications.
The conference disclosed that more than 1.1 million enterprises and individuals are currently using Volcano Ark large model services, with 200 enterprises achieving annual token call volumes exceeding 1 trillion, doubling within six months, covering industries including internet, manufacturing, finance, and automotive.
