豆包 API

Seed 2.0 lineup

Volcengine Ark open API—Seed 2.0, Seedance, and Seedream stack with per-token or per-call billing

Token API

Core models

Doubao-Seed-2.0-ProDoubao-Seed-2.0-LiteDoubao-Seed-2.0-MiniDoubao-Seed-2.0-CodeDoubao-Seed-1.8Doubao-Seed-CharacterDoubao-Seed-CodeDoubao-Seed-1.6Doubao-Seed-1.6-LiteDoubao-Seed-1.6-FlashDoubao-Seed-1.6-VisionDoubao-Seed-TranslationDoubao-1.5-Pro-32kDoubao-1.5-Lite-32kDoubao-1.5-Vision-ProDoubao-Pro-32kDoubao-Seedance-2.0Doubao-Seedance-2.0-FastDoubao-Seedance-2.0-MiniDoubao-Seedance-1.5-ProDoubao-Seedance-1.0-ProDoubao-Seedance-1.0-Pro-FastDoubao-Seedream-5.0-LiteDoubao-Seedream-4.5Doubao-Seedream-4.0Doubao-Seed3D-2.0Doubao-Embedding-Vision

Doubao-Seed-2.0-Pro

Seed 2.0 flagship—≤32K band ¥3.2/¥16 per 1M tokens for complex reasoning and agents.

Doubao-Seed-2.0-Lite

Seed 2.0 Lite workhorse—≤32K input ¥0.6/¥3.6 per 1M with cache-hit discounts.

Doubao-Seed-2.0-Mini

Seed 2.0 Mini—≤32K input ¥0.2/¥2 per 1M, lowest listed Ark API entry.

Doubao-Seed-2.0-Code

Seed 2.0 Code—≤32K input ¥3.2/¥16 per 1M with strong coding and frontend skills.

Doubao-Seed-1.8

Seed 1.8 general model—≤32K input ¥0.8, output ¥2–8/1M by output length band.

See the official site for more models

Additional core model names still appear above, with full details on the latest official page.

Open official site

Plan details

Doubao-Seed-2.0-Mini

Lowest list

Input

¥0.2

Output

¥2

Official

Doubao-Seed-2.0-Mini

Lowest list

Input

¥0.2

Output

¥2

Official

Usage

doubao-seed-2.0-mini ≤32K input ¥0.2, output ¥2/1M—the lowest listed LLM rate for high-volume light workloads.

Models

Supports audio input bands (¥3/1M audio input); longer context bands cost more—plan by real prompt length.

Highlights

If Lite meets quality, Mini maximizes throughput; route complex reasoning to Pro or Code.
Same capability inside Coding Plan uses subscription request quotas—not this open API path.

Best for

High-volume light calls and cost-sensitive baseline traffic

Doubao-Seed-2.0-Lite

Mainstream valueRecommended

Input

¥0.6

Output

¥3.6

Official

Doubao-Seed-2.0-Lite

Mainstream valueRecommended

Input

¥0.6

Output

¥3.6

Official

Usage

doubao-seed-2.0-lite ≤32K input ¥0.6, output ¥3.6/1M, cache-hit input ¥0.12—for daily apps and agents at scale.

Models

Higher rates for (32,128]K and (128,256]K input bands; audio input billed separately.

Highlights

Default layer for most production traffic—escalate to Pro when quality demands rise.
Batch inference ~50% of online—evaluate batch channel for steady volume.

Best for

Teams running daily AI apps and agents with cost sensitivity

Doubao-Seed-2.0-Pro

Seed 2.0 flagship

Input

¥3.2

Output

¥16

Official

Doubao-Seed-2.0-Pro

Seed 2.0 flagship

Input

¥3.2

Output

¥16

Official

Usage

doubao-seed-2.0-pro ≤32K input ¥3.2, output ¥16/1M—top choice for complex reasoning, long agents, and production Q&A.

Models

Unit rates rise sharply above 32K input; low-latency online inference is 3× regular online rates.

Highlights

Better as key-path escalation than default routing for all traffic.
Same model name exists in Coding Plan subscriptions—but different keys and billing.

Best for

Complex agents, production Q&A, and critical business flows

Doubao-Seed-2.0-Code

Coding

Input

¥3.2

Output

¥16

Official

Doubao-Seed-2.0-Code

Coding

Input

¥3.2

Output

¥16

Official

Usage

doubao-seed-2.0-code ≤32K input ¥3.2, output ¥16/1M—for IDE assist, code review, and engineering agents.

Models

Strong multimodal vision and frontend generation—heavy individual coding may compare Coding Plan subscription economics.

Highlights

Versus doubao-seed-code (¥1.2/¥8), 2.0-Code is stronger but pricier.
Open API path is fully separate from Coding Plan request-based subscription.

Best for

Coding assistants, IDE integrations, and engineering agent teams

Notes

entryPrice uses doubao-seed-2.0-mini ≤32K lowest band; longer context and stronger models raise unit rates—estimate on real traffic before launch.
The pricing page lists hosted third-party rates (GLM, DeepSeek, etc.), but Doubao API `models` excludes them—see plans.json for subscription bundles.
Batch inference on some Seed models is ~50% of online rates; TPM packs, model units, and web plugins are separate billing lines—not separate tiers here.
Third-party 3D models like Hyper3D-Gen2 and Hitem3D appear on the pricing page but are not Doubao-owned API catalog entries.

Supported coding tools

OpenAI-compatible APIArkTRAEBatch inferenceContext cache

Pricing and model data sourced from official vendor websites

FAQ

General·6

General

6 条