Doubao Seed 2.0 API

Doubao Seed 2.0 is ByteDance's latest model family for production text workloads. EvoLink gives you one OpenAI-compatible endpoint for Pro, Lite, Mini, and Code, so you can choose the right model for quality, cost, latency, or coding tasks without changing providers.

Model Type:

Seed 2.0 Pro Seed 2.0 Lite Seed 2.0 Mini ✓Seed 2.0 Code

Price:

$0.444 - 1.333(~ 32 - 96 credits) per 1M input tokens; $2.222 - 6.667(~ 160 - 480 credits) per 1M output tokens

$0.089(~ 6.4 credits) per 1M cache read tokens

Cache storage charged separately per hour.

Highest stability with guaranteed 99.9% uptime. Recommended for production environments.

Use the same API endpoint for all versions. Only the model parameter differs.

Doubao Seed 2.0 Code API pricing and access

Use Doubao Seed 2.0 Code through one EvoLink API key. This page focuses on the Code variant for programming workflows, coding agents, and software engineering tasks that need stronger code-centric output.

PRICING

PLAN	CONTEXT WINDOW	MAX OUTPUT	INPUT	OUTPUT	CACHE READ
Doubao Seed 2.0 Code	256,000	128,000	≤32.0K$0.444 (32 Credits) ≤128.0K$0.667 (48 Credits) ≤256.0K$1.333 (96 Credits)	≤32.0K$2.222 (160 Credits) ≤128.0K$3.333 (240 Credits) ≤256.0K$6.667 (480 Credits)	≤32.0K$0.089 (6.4 Credits) ≤128.0K$0.089 (6.4 Credits) ≤256.0K$0.089 (6.4 Credits)
Cache Storage Cached prompt tokens billed by storage duration					$0.002/1M tokens/hour (0.17 Credits)

Pricing Note: Prices show both USD and Credits. Units default to / 1M tokens unless noted separately.

Cache Hit: Price applies to cached prompt tokens.

What can you build with Doubao Seed 2.0 Code?

Coding Agents and Developer Tools

Use Seed 2.0 Code for agent workflows that generate, inspect, or revise code inside developer platforms, automation tools, and internal engineering systems.

Build coding agents

Use-case showcase of Doubao Seed 2.0 feature 1

Code Review and Refactoring

Seed 2.0 Code is a better fit for reviewing diffs, explaining code paths, proposing refactors, and supporting engineering workflows where output quality matters more than generic chat behavior.

Improve code review

Use-case showcase of Doubao Seed 2.0 feature 2

Debugging and Technical Docs

Use Code for debugging help, stack-trace interpretation, test suggestions, and technical documentation across multi-language software projects.

Support debugging

Use-case showcase of Doubao Seed 2.0 feature 3

Why teams choose Doubao Seed 2.0 Code

Seed 2.0 Code is the right variant when your API purchase is driven by software engineering output, coding agents, and developer productivity rather than general chat alone.

Programming-focused output

Code gives you a more specialized profile for software engineering tasks than a general-purpose text model.

Unified access for coding stacks

Keep one EvoLink integration while routing engineering workloads to the Code variant and other traffic to the rest of the Seed 2.0 family.

Useful for developer operations

Use Code in internal tooling, CI helpers, code review assistants, and engineering support flows.

How to integrate Doubao Seed 2.0 Code API

Use the EvoLink API with the Code model enum. Migration is usually a base URL swap plus `doubao-seed-2.0-code` in your existing OpenAI-compatible workflow.

Step 1 — Authenticate

Create an EvoLink API key and send requests with Bearer token authentication from your app, agent, or backend service.

Step 2 — Use the Code model ID

Set model to `doubao-seed-2.0-code` when you want the Code variant on EvoLink.

Step 3 — Tune outputs

Adjust temperature, top_p, max_tokens, stream, and other parameters. Prompt-length tiers and cache billing are applied automatically in the API pricing model.

View API Docs

Core Doubao Seed 2.0 Code API capabilities

Model facts for the programming-focused Seed 2.0 variant, plus EvoLink pricing and access signals

Coding

Programming-focused model profile

Code is optimized for software engineering output such as code generation, debugging, review, refactoring, and technical explanation.

Context

256K Context Window

Use Code with long files, larger repositories, and complex engineering prompts that need wider context.

Pricing

Prompt-Length Pricing

Code pricing scales by prompt length, making it easier to evaluate software engineering workloads against real API usage.

Caching

Cache Hit Billing

Reduce repeated prompt cost in agent scaffolds, engineering templates, and reusable coding instructions.

Compatibility

OpenAI-Compatible Access

Use Code through one EvoLink endpoint and plug it into existing coding agents or OpenAI-compatible tooling.

Use Cases

Developer workflow fit

Code is best suited for engineering copilots, review tools, debugging assistants, and internal developer platforms.

Doubao Seed 2.0 Code API FAQs

Everything you need to know about the product and billing.

Doubao Seed 2.0 Code API is the programming-focused model in the Seed 2.0 family. On EvoLink it is available through the same OpenAI-compatible API gateway as the other Seed 2.0 variants.

Choose Code when your workload is centered on software engineering: code generation, review, debugging, refactoring, technical documentation, and coding agents.

Seed 2.0 uses length-based pricing tiers for input and output. Requests within 32K prompt length are priced lower than 128K or 256K tiers, so cost scales with how much context you actually consume. On EvoLink, the pricing section below is the canonical place to compare costs by variant.

Seed 2.0 supports cache hit billing. When the same system prompts or prompt prefixes are reused, cached tokens are billed below standard input rates. That makes a visible difference in production workloads with repeated prompt scaffolding, agent prefixes, or stable enterprise instructions.

Use these model enums in your API requests: doubao-seed-2.0-pro, doubao-seed-2.0-lite, doubao-seed-2.0-mini, and doubao-seed-2.0-code. EvoLink keeps access unified behind one API gateway, so variant selection happens through the model field rather than a separate vendor integration.

Yes. EvoLink exposes an OpenAI-compatible API endpoint for Seed 2.0. In most integrations, migration means changing the base URL, using your EvoLink API key, and selecting the right Seed 2.0 model enum.

Seed 2.0 models support a maximum output of 128,000 tokens (128K) per request. The context window supports up to 256K tokens for input.

Pro fits premium assistants, long-form generation, and higher-quality general tasks. Lite fits cost-aware chat and content workloads. Mini fits real-time responses, routing, classification, and large-scale throughput. Code fits software engineering, code review, debugging, and developer tooling.