HuoShanYinQing

Overview

Volcengine is ByteDance's enterprise-grade AI cloud platform, built to deliver a comprehensive suite of models, tools, applications, and cloud infrastructure for one-stop enterprise AI development. As a leading AI cloud in China, Volcengine accounts for 49.5% of the country's large language model public cloud market call volume, providing organizations with a robust, production-ready foundation to build, deploy, and scale AI-powered products and services. The platform is purpose-built for the Agent era, ensuring that AI agents can go live faster, run more stably, and scale more easily.

At the core of the platform is Volcano Ark (火山方舟), a one-stop large model service platform that enables enterprises to access, fine-tune, and deploy large language models through unified APIs. The platform supports the full AI lifecycle, including model inference, model training, and application development. Developers can get started quickly using the Ark CLI via npm (@volcengine/ark-cli) to integrate model capabilities directly into terminal environments or AI agents. Complementary products include AgentKit, an enterprise-level AI Agent platform; Coze (扣子), a next-generation enterprise AI Agent productivity engine; HiAgent, a "1+N+X" one-stop intelligent agent workstation; TRAE CN, an AI-native software engineering assistant; ArkClaw Enterprise, an intelligent productivity system for AI-powered employees; and a cloud-native machine learning platform for enterprise model training and management.

Volcengine offers a diverse portfolio of self-developed large models covering text, image, video, and speech modalities. The Doubao large model family includes the Evolving model, which is continuously updated and focused on Coding and Agent scenarios, priced at ¥6.00 per million input tokens and ¥30.00 per million output tokens. The Seedance 2.5 video generation model delivers 30-second long-form narrative generation with expanded full-modal reference support including video input, priced at ¥42 per million tokens with video input and ¥70 per million tokens without. The Doubao Image Creation Model 5.0 enables more intelligent and controllable creative generation with built-in web retrieval, priced at ¥0.22 per image for both text-to-image and image-to-image tasks. The Doubao Speech Recognition Model 2.0 provides accurate hybrid-language recognition and deep audio analysis, priced at ¥0.8 per hour for recording file recognition and ¥1.0 per hour for streaming speech recognition.

Beyond AI models, Volcengine provides a complete AI-native cloud infrastructure, including elastic GPU cloud servers, object storage with 10EB-level data scale, cloud-native MySQL databases, containers, middleware, big data analytics, video and edge computing, security, and enterprise services. The platform has earned more than 20 certifications and recognitions from authoritative organizations including Gartner, IDC, and CAICT, covering model trustworthiness, intelligent agent controllability, and security operations. With 7x24-hour professional engineering support, second-level service response, a customer-value-first service philosophy, and end-to-end security assurance, Volcengine empowers enterprises across automotive, financial, entertainment, healthcare, media, smart tourism, and consumer industries to accelerate business innovation and digital transformation in the AI-driven era.

Core capabilities

Volcano Ark (One-Stop LLM Service Platform)

Unified API access to large language models covering the full AI lifecycle of inference, training, and application development, with CLI support via the npm package @volcengine/ark-cli.

Doubao Large Model Family

ByteDance's self-developed foundation models, including the continuously evolving Evolving model optimized for Coding and Agent scenarios with competitive token-based pricing.

Seedance 2.5 Video Generation Model

30-second long-form narrative video generation with full-modal reference expansion and optional video input support for richer multimodal understanding.

Doubao Image Creation Model 5.0

Intelligent and controllable image generation with built-in web retrieval, supporting both text-to-image and image-to-image at ¥0.22 per image.

Doubao Speech Recognition Model 2.0

Hybrid-language accurate speech recognition with deep audio analysis, supporting both recording file recognition and streaming speech recognition.

AgentKit (Enterprise AI Agent Platform)

Enterprise-level platform for building, deploying, and governing AI agents in production environments.

Coze (AI Agent Productivity Engine)

Next-generation enterprise AI Agent engine for creating AI-powered workflows, chatbots, and intelligent employee assistants.

HiAgent (Intelligent Agent Workstation)

A "1+N+X" one-stop intelligent agent workstation for constructing and orchestrating enterprise-grade intelligent agents.

TRAE CN (AI Software Engineer)

AI-native software engineering tool enabling Vibe Coding workflows and full-cycle AI-assisted code development.

AI-Native Cloud Infrastructure

GPU cloud servers, 10EB-level object storage, cloud MySQL databases, and a cloud-native machine learning platform with 20+ industry certifications from Gartner, IDC, and CAICT.

7x24 Enterprise Support and Security

Round-the-clock professional engineering services, second-level response, and a transparent, trusted cloud security framework.

Use cases

Building enterprise AI agents and intelligent chatbots for customer service, internal knowledge management, and employee productivity.

Vibe Coding and AI-assisted software development using TRAE CN for end-to-end code generation and review.

Generating creative advertising content and marketing assets with the Doubao Image Creation Model 5.0.

Producing AI short dramas and long-form narrative video content using the Seedance 2.5 video generation model.

Deploying large language models for coding assistance and intelligent agent workflows via Volcano Ark's unified API and CLI.

Transcribing and analyzing audio content with the Doubao Speech Recognition Model 2.0 for recordings and streaming media.

Running enterprise-scale machine learning training and inference on GPU cloud infrastructure with the cloud-native ML platform.