Volcengine

Doubao Seed and multi-model aggregation — ByteDance's cloud AI platform

Website·Docs·Pricing·Sign Up

API Configuration

API Type	`openai-completions`
Base URL	`https://ark.cn-beijing.volces.com/api/v3`
Auth Mode	api_key
Auth Keys	`VOLCENGINE_API_KEY` — API Key — obtain from https://console.volcengine.com/
Regions	cn
Default Model	`doubao-seed-1-8-251228`
Status	active

Pricing

Pay-as-you-go per token (CNY) with tiered pricing. See volcengine.com docs for per-model rates.

Getting Started

Sign up at console.volcengine.com, activate Doubao/Ark service, then create an API key.

Models (15)

Model	Context	Max Output	Pricing	Features
DeepSeek V3.2 (via Volcengine) `deepseek-v3-2-251201` DeepSeek V3.2 hosted on Volcengine with 128K context	128K	4K	Tiered (CNY): Input Y2/M (<=32K) / Y4/M (<=128K), output Y3/M (<=32K) / Y6/M (<=128K).
Kimi K2.5 (via Volcengine) `kimi-k2-5-260127` Moonshot's Kimi K2.5 hosted on Volcengine	256K	4K	CNY: Input Y4/M, output Y16/M.
Doubao-Seed-Evolving `doubao-seed-evolving` Seed's latest coding and agent model; unified Model ID with 1M context	1.0M	262K	Pay-as-you-go: ¥6 per million input tokens, ¥30 per million output tokens. Also included in Ark plan subscriptions.
Doubao-Seed-2.1-pro `doubao-seed-2.1-pro` Production-grade model upgraded for coding, agent and multimodal tasks	262K	262K	Pay-as-you-go: ¥6 per million input tokens, ¥30 per million output tokens.
Doubao-Seed-2.1-turbo `doubao-seed-2.1-turbo` Cost/quality balanced model upgraded for coding, agent and multimodal tasks	262K	262K	Pay-as-you-go: ¥3 per million input tokens, ¥15 per million output tokens. Also included in Ark plan subscriptions.
Doubao-Seed-2.0-lite `doubao-seed-2.0-lite` Doubao's first omni-modal understanding model — audio, image, video and text	262K	131K	Pay-as-you-go (CNY, reasoning tiers): input ¥0.6–1.8 per million tokens, output ¥3.6–10.8 per million tokens.
GLM-5.2 `glm-5.2` GLM-5.2 with a large coding upgrade for complex development	1.0M	131K	Pay-as-you-go: ¥8 per million input tokens, ¥28 per million output tokens.
GLM-5.3 `glm-5.3` GLM-5.3 served through the Ark plan endpoints, with 1M context	1.0M	-	Included in the Ark Coding Plan / Agent Plan subscription.
GLM-5.3-Flash `glm-5.3-flash` Faster GLM-5.3 variant served through the Ark plan endpoints	1.0M	131K	Included in the Ark Coding Plan / Agent Plan subscription.
DeepSeek-V4-pro `deepseek-v4-pro` DeepSeek V4 flagship hybrid reasoning model	1.0M	393K	Pay-as-you-go: ¥12 per million input tokens, ¥24 per million output tokens. Also included in Ark plan subscriptions.
DeepSeek-V4-flash `deepseek-v4-flash` DeepSeek V4 fast hybrid reasoning model	1.0M	393K	Pay-as-you-go: ¥1 per million input tokens, ¥2 per million output tokens. Also included in Ark plan subscriptions.
DeepSeek-V4.1-Flash `deepseek-v4.1-flash` DeepSeek V4.1 Flash served through the Ark Agent Plan	1.0M	393K	Included in the Ark Agent Plan subscription.
Kimi K3 `kimi-k3` Kimi K3 with 1M context, available on selected Agent Plan tiers	1.0M	-	Included in the Ark Agent Plan subscription (selected tiers).
Kimi K2.8 Preview `kimi-k2.8-preview` Kimi K2.8 preview served through the Ark Agent Plan	1.0M	-	Included in the Ark Agent Plan subscription; preview model with limited-time discounts.	(preview)
MiniMax M3 `minimax-m3` MiniMax M3 served through Ark. Ark documents a 1024k context window and a 128k maximum output for it, and does not list image input on its model table, so this record is text-only.	1.0M	131K	Included in the Ark Coding Plan and Agent Plan subscriptions; see the plan pages for deduction ratios.

Last verified: 2026-09-21