兔子API
provider
ByteDance Doubao
Direct answerDirect Ark access uses a project API key and https://ark.cn-beijing.volces.com/api/v3. Preset models normally use a model ID, custom deployments use an endpoint ID, and this gateway uses its own token and enabled model name.
Updated · Reviewed
Direct access versus this gateway
Ark's direct data-plane base is https://ark.cn-beijing.volces.com/api/v3 and uses a project API key. Its control plane uses ark.cn-beijing.volcengineapi.com and is not the inference address. Gateway calls use this site's base URL and Bearer token, and this page currently declares only /v1/chat/completions. Ark recommends evaluating Responses for new direct integrations, but that does not expose the same gateway route.
Model ID, endpoint ID, and gateway model name
A console display name is for reading. Preset model calls normally use a model ID, such as doubao-seed-2-1-pro-260628 in the official quickstart; a custom inference deployment uses an endpoint ID. Gateway clients use a name enabled in the gateway model list because administrators may map it to either official identifier. Do not copy these three values interchangeably.
Authentication, projects, and protocol boundaries
An Ark API key belongs to one project and can be restricted by model, custom endpoint, and source IP. Direct OpenAI-compatible extensions often go in extra_body, while capabilities such as embeddings may require a native protocol. Media generation has separate task workflows and cannot be called with Chat merely because it shares the Doubao brand.
Streaming, tools, and errors
Official Chat and Responses use different SSE event shapes; a Responses stream cannot be parsed as Chat delta. Web search, knowledge bases, MCP, and function tools are model- and protocol-specific, and applications must validate tool arguments and side effects. A stream may carry an error after HTTP 200, so keep reading error.type and the request ID.
Cost safety and production checklist
Coding Plan uses a separate base URL, and the pay-as-you-go address can create out-of-plan charges. Set output caps, concurrency, timeouts, bounded retries, and account alerts, and verify tool, cache, and media pricing. Before launch, check region, project, key scope, model or endpoint identifier, protocol support, moderation, and current pricing.
Reviewed provider catalog
Latest models by capability
These are provider-catalog models, not a promise of availability on this site. Confirm callable IDs, endpoints, and pricing in the model marketplace.
Text, reasoning, and code
-
doubao-seed-evolvingStableDoubao's continuously evolving agent and coding model with a fixed ID that tracks the strongest production version.
-
doubao-seed-2.1-proStableThe flagship Doubao Seed 2.1 model for complex coding, long-running agents, and multimodal reasoning.
-
doubao-seed-2.1-turboStableThe balanced Doubao Seed 2.1 model for production workloads at scale.
Image generation and editing
-
doubao-seedream-5-0-260128StableThe high-quality Doubao Seedream 5.0 image generation and editing model.
-
doubao-seedream-5-0-lite-260128StableThe lightweight, high-throughput image model in the Seedream 5.0 family.
Video generation and editing
-
doubao-seedance-2-0-260128StableThe high-quality Doubao Seedance 2.0 video generation model.
-
doubao-seedance-2-0-fast-260128StableThe faster video generation model in the Seedance 2.0 family.
Embeddings, retrieval, and reranking
-
doubao-embedding-visionStableDoubao's stable multimodal embedding alias for text, image, and video retrieval.
-
doubao-embedding-vision-250615StableA pinned, reproducible snapshot of Doubao's multimodal embedding model.
Use cases
- General agents and reasoning
- Multimodal understanding
- Image and video generation
API protocols
/v1/chat/completions
FAQ
What is the difference between direct Ark access and this gateway?
Direct access uses Ark's data-plane base https://ark.cn-beijing.volces.com/api/v3, a project API key, and an official model identifier. This gateway uses its own base URL, Bearer token, and /v1/chat/completions declared here. Official Responses support does not mean the gateway exposes /v1/responses.
Should model contain a display name, model ID, or endpoint ID?
A console display name is for humans. Preset models normally use an official model ID such as the documented doubao-seed-2-1-pro-260628; a custom inference deployment uses an endpoint ID. Through this gateway, use an enabled gateway model name because an administrator may map it.
Why can authentication fail when the key looks correct?
An Ark API key belongs to a project and can be restricted by model, custom endpoint, and source IP; it cannot be reused across projects. Moving an endpoint can invalidate the old key, and the data-plane domain must not be confused with the volcengineapi.com control plane.
How do I avoid surprise Coding Plan or streaming charges?
Coding Plan has a separate base URL; using a pay-as-you-go base can bypass the plan and create extra charges. Cap output, concurrency, and retries, and verify tool and cache billing. SSE can also carry a business error, so parse error.type instead of trusting HTTP 200 alone.
Related guides
Official sources
- Volcengine Ark Beginner Quickstart Official
- Volcengine Ark Model Call Quickstart Official
- Volcengine Ark API Key Configuration Official
- Volcengine Ark Base URLs and Authentication Official
- Volcengine Ark OpenAI SDK Compatibility Official
- Volcengine Ark Error Codes Official