Skip to main content
Doubao / VolcArk is ByteDance’s collection of large models made available through Fhddos via a unified OpenAI-style interface. You get access to state-of-the-art text and multimodal chat, image generation, video synthesis, and text-to-speech — all authenticated with your Fhddos API key, no VolcArk console setup required.
You only need a Fhddos API key (oh-xxxxxxxx) and the API model name listed in this page. Fhddos automatically maps these short names to internal versioned model IDs — you never need to use long dated names like doubao-seed-1-6-251015.

​
What You Can Do

Text & Multimodal Chat

Use Doubao Seed, DeepSeek, Kimi K2, and GLM models via /v1/chat/completions or /v1/responses for conversation, reasoning, tool calling, and vision tasks.

Image Generation

Generate images from text, edit existing images, or compose multi-image scenes with Seedream and SeedEdit models via /v1/images/generations.

Video Generation

Create text-to-video and image-to-video outputs with Seedance models using either the OpenAI-style /v1/videos interface or the native VolcArk endpoint.

Text-to-Speech

Synthesize speech with Volcengine OpenSpeech via /volcark/openspeech/*, supporting V1/V3 HTTP and WebSocket protocols.

​
Authentication & Base URL

All Doubao API calls use the same base URL and Bearer authentication: 
export BASE_URL="https://aiapi.fhddos.com/v1"
export TOKEN="oh-xxxxxxxxxxxxxxxx"

Authorization: Bearer <TOKEN>
Content-Type: application/json

​
Available Models

​
Text & Multimodal Models

Use these with /v1/chat/completions or /v1/responses:
API Model NameBest For
doubao-seed-1.8General Chinese/English, deep reasoning
doubao-seed-1.6Stable, compatible general chat
doubao-seed-1.6-liteLow-cost version
doubao-seed-1.6-flashSpeed-sensitive scenarios
doubao-seed-1.6-visionImage understanding / vision
doubao-seed-codeCode generation and engineering
doubao-seed-translationTranslation and multilingual writing
deepseek-v3.2Advanced reasoning
deepseek-v3.1Advanced reasoning (stable)
deepseek-v3Open-source compatible reasoning
deepseek-r1-arkReasoning-enhanced chain-of-thought
kimi-k2Chain-of-thought reasoning
glm-4.7General chat (requires admin enablement)
Model names are case-insensitive within Fhddos, but use the lowercase API names from this table to avoid confusion. If you see an InvalidEndpointOrModel.NotFound error, check that you haven’t accidentally used the long internal version ID (e.g. doubao-seed-1-6-251015) instead of the short API name.

​
Model Name → Internal ID Mapping

Fhddos transparently maps short API names to versioned upstream IDs. This table is for reference only — you never need to use these IDs:
API Model NameUnderlying Model ID (example)
doubao-seed-1.8doubao-seed-1-8-251228
doubao-seed-1.6doubao-seed-1-6-251015
doubao-seed-1.6-litedoubao-seed-1-6-lite-251015
doubao-seed-1.6-flashdoubao-seed-1-6-flash-250828
doubao-seed-1.6-visiondoubao-seed-1-6-vision-250815
doubao-seed-codedoubao-seed-code-preview-251028
doubao-seed-translationdoubao-seed-translation-250915
deepseek-v3.2deepseek-v3-2-251201
deepseek-r1-arkdeepseek-r1-250528
kimi-k2kimi-k2-thinking-251104

​
Image Generation Models

Use these with /v1/images/generations:
API Model NameCapability
doubao-seedream-4.5Text-to-image, image-to-image, multi-image, sequential generation, streaming
doubao-seedream-4.5-nSame as above — per-image billing variant
doubao-seedream-4.0Text-to-image, image-to-image, multi-image, sequential generation, streaming
doubao-seedream-4.0-nSame as above — per-image billing variant
doubao-seedream-3.0-t2iText-to-image with seed and guidance_scale control
doubao-seededit-3.0-i2iImage-to-image editing

​
Video Generation Models

Use these with /v1/videos (OpenAI style) or /volcark/api/v3/contents/generations/tasks (native):
API Model NameType
doubao-seedance-1-5-proText-to-video (T2V), audio support
doubao-seedance-1-0-proText-to-video (T2V)
doubao-seedance-1-0-pro-fastText-to-video, faster variant
doubao-seedance-1-0-lite-t2vText-to-video, lightweight
doubao-seedance-1-0-lite-i2vImage-to-video (I2V)

​
How to Choose a Model

Start with doubao-seed-1.8 for the best overall quality. Use doubao-seed-1.6 for stability, doubao-seed-1.6-lite to reduce cost, or doubao-seed-1.6-flash when latency matters most.
Use doubao-seed-1.8 (or its alias doubao-seed-1.6-thinking) for Doubao deep reasoning. For alternatives, try deepseek-v3.2, deepseek-r1-ark, or kimi-k2.
Use doubao-seed-code as your primary choice. deepseek-v3.1 and deepseek-v3 are strong alternatives with a larger open-source ecosystem.
Use doubao-seed-translation for dedicated translation tasks. For translation that requires deep context understanding, doubao-seed-1.6 also works well.
Use doubao-seed-1.6-vision. Pass images alongside text in messages.content using the image_url type, exactly as you would with OpenAI Vision models.

​
What You Don’t Need to Worry About

As an API caller, you do not need to:
  • Open or configure models in the VolcArk console
  • Manage VolcArk base URLs, credentials, or request signing
  • Decide how channels route or load-balance requests
  • Track model version upgrades — Fhddos handles these automatically in the background
Your only steps are: get a Fhddos API key, pick a model name from this page, and call /v1/chat/completions, /v1/responses, /v1/images/generations, or /v1/videos.