Doubao-Seed-2.0-lite

doubao-seed-2-0-lite-260215

Get API keyAPI reference

A balanced model for high-frequency enterprise scenarios that balances performance and cost, with overall capabilities surpassing the previous generation Doubao-Seed-1.8. It is well-suited for production tasks such as unstructured information processing, content creation, search recommendation, and data analysis, supporting long context, multi-source information fusion, multi-step instruction execution, and high-fidelity structured output. It significantly optimizes costs while ensuring stable performance.

Context window1.0M

Provider豆包

Released2026/04/27

Playground

Pricing

The same model is available through multiple service channels — choose based on latency, reliability and cost.

Prices in $ / 1M tokens

To pick a channel, add a provider field to the request body, for example "provider": { "channel": "direct" }. Valid values are direct / stable / economical; omit it to use the default channel.

Direct

Direct upstream connection — best when you need native behavior and the full context window.

Input context	Input	Output	Cache read
≤ 32K	0.09/M	0.53/M	0.02/M
32K – 128K	0.13/M	0.79/M	0.03/M
> 128K	0.26/M	1.59/M	0.05/M

Overview

Input

Text Image

Output

Text

Doubao-Seed-2.0-lite API: An Omni-Modal Model Balancing Performance and Cost

Doubao-Seed-2.0-lite is a balanced model launched by ByteDance for high-frequency enterprise scenarios, balancing performance and cost, with overall capabilities surpassing the previous generation Doubao-Seed-1.8. It is well-suited for production workloads such as unstructured information processing, content creation, search and recommendation, and data analysis, supporting long context, multi-source information fusion, multi-step instruction execution, and high-fidelity structured output.

The new version from May 2026 makes it the first omni-modal understanding model in the Doubao large model family—supporting native unified understanding of video, image, audio, and text. It significantly surpasses the Pro version released in February on advanced subject reasoning benchmarks such as physics (HiPhO) and medicine (MedXpertQA), and reaches SOTA levels in fine-grained perception (BabyVision, WorldVQA) and embodied understanding (ERQA).

SeaWhale AI provides Doubao-Seed-2.0-lite through an OpenAI-compatible interface, supporting multimodal input, tool calling, and streaming output.

Get the API key · Model ID: doubao-seed-2-0-lite-260215

---

Why Choose Doubao-Seed-2.0-lite

• The first omni-modal understanding model in the Doubao family — native unified understanding of video, image, audio, and text
• Advanced subject reasoning surpasses the Pro version — significant lead in physics (HiPhO) and medicine (MedXpertQA)
• SOTA in fine-grained perception and embodied understanding — leading in BabyVision, WorldVQA, and ERQA
• Balances performance and cost — designed for high-frequency enterprise scenarios, economically viable for large-scale deployment
• 1 million token context — long context and multi-source information fusion
• High-fidelity structured output — reliable integration with production workflows

---

Core Capabilities

01 Native Omni-Modal Understanding

This is the biggest upgrade in the May version: video, image, audio, and text are natively and uniformly understood within the same model, rather than stitching together multiple models. This makes cross-modal associative reasoning possible.

• Video content understanding and temporal reasoning
• Audio understanding and transcription
• Unified parsing of mixed image-text materials

02 Advanced Subject Reasoning

On advanced subject reasoning benchmarks such as physics (HiPhO) and medicine (MedXpertQA), the new lite version significantly surpasses the Pro version released in February, demonstrating that the capability improvements in this version are substantial.

• Physics and scientific reasoning
• Medical professional problem analysis
• Interdisciplinary knowledge synthesis

03 Fine-Grained Perception and Embodied Understanding

Reaches SOTA levels on fine-grained perception benchmarks such as BabyVision and WorldVQA and the ERQA embodied understanding benchmark, making it suitable for scenarios requiring precise visual judgment.

• Fine-grained visual difference recognition
• Spatial relationship and physical reasoning
• Embodied intelligence related scenarios

04 Production Workloads

lite's core focus remains high-frequency production work: unstructured information processing, content creation, search and recommendation, and data analysis—these scenarios require stability, speed, and low cost.

• Unstructured information processing
• Content creation and rewriting
• Search and recommendation, and data analysis

---

Best Use Cases

Use Case	Description
Omni-modal content understanding	Unified processing of video, audio, image, and text
High-frequency information processing	Extraction and organization of large-scale unstructured data
Content creation	Batch generation of copy, creative assets, and marketing content
Search and recommendation	Relevance judgment and content understanding
Subject reasoning	Professional domain problems such as physics and medicine
Fine-grained visual judgment	Scenarios requiring precise visual difference recognition

---

Differences Between Doubao-Seed-2.0-lite and the Pro Version

Capability	Doubao-Seed-2.0-lite	Doubao-Seed-2.0-pro
Model ID	doubao-seed-2-0-lite-260215	doubao-seed-2-0-pro-260215
Positioning	Balanced for high-frequency enterprise scenarios	Flagship all-purpose general model
Context window	1 million tokens	1 million tokens
Max output	65.5K tokens	65.5K tokens
Modalities	Omni-modal (video/image/audio/text)	Text, image
Advanced subject reasoning	Significantly surpasses the February Pro version	Strong
Focus	Balance of performance and cost	Complex reasoning and long-chain execution

The Doubao series is priced in tiers by input length (32K / 128K / longer); please refer to the real-time price card at the top of the page for exact pricing.

---

How to Use the Doubao-Seed-2.0-lite API

1. Create a SeaWhale AI API key Generate a key in the console and top up your balance.

2. Try it first on multimodal tasks The lite version's omni-modal understanding capability is the biggest differentiator of this generation. For video and audio related scenarios, you can use it directly without stitching together multiple specialized models.

3. Call the API

curl -X POST https://api.atalk-ai.com/v1/chat/completions \
  -H 'Authorization: Bearer YOUR_API_KEY' \
  -H 'Content-Type: application/json' \
  -d '{
    "model": "doubao-seed-2-0-lite-260215",
    "messages": [{
      "role": "user",
      "content": [
        {"type": "image_url", "image_url": {"url": "https://example.com/chart.png"}},
        {"type": "text", "text": "Read the values for each quarter in this chart and describe the trend."}
      ]
    }],
    "stream": true
  }'

---

Frequently Asked Questions

Why does the lite version outperform the Pro version on some benchmarks? lite received a new version upgrade in May 2026, becoming the first omni-modal understanding model in the Doubao family. On advanced subject reasoning such as physics and medicine, the new lite version significantly surpasses the Pro version released in February.

What exactly does "omni-modal" include? Native unified understanding of video, image, audio, and text—four modalities processed within the same model, rather than stitching together multiple specialized models.

When should you still use Pro? When you need the most complex multi-step planning, long-chain execution, and simultaneous satisfaction of multiple constraints, the Pro version is still more suitable.

What are the context and output limits? 1 million token context, maximum output of 65,500 tokens.

Why are prices divided into three tiers? The Doubao series is priced in tiers by input token length (32K / 128K / longer). The unit price for short inputs is significantly lower, making context length control an effective way to reduce costs.

Is it suitable for large-scale deployment? Yes. lite is precisely the tier designed for high-frequency enterprise scenarios, significantly optimizing costs while ensuring stable performance.

---

Why Choose SeaWhale AI for the Doubao-Seed-2.0-lite API

• No Volcano Engine account required — call directly with one API key
• OpenAI-compatible interface — multimodal request format consistent with OpenAI
• Unified omni-modal entry point — video, audio, image, and text share one set of APIs
• Settled in RMB — domestic models priced in RMB

API

API integration

Model IDUse this value as the model in inference requests

doubao-seed-2-0-lite-260215

API KeyBearer token used to authenticate inference requests

Manage API keys

Base URLOpenAI compatible · /chat/completions

https://api.atalk-ai.com/v2/

provider OptionalSelects a service channel; omit it and the system picks the default

"provider": { "channel": "direct" }

doubao-seed-2-0-lite-260215 usage examples

SeaWhale AI is compatible with the OpenAI API protocol, so you can call it with the OpenAI SDK or plain HTTP requests. Streaming is enabled by default.

About the provider parameter (optional, a SeaWhale AI extension): most models are served over several channels that differ slightly in price and reliability. Add a provider field to the request body to pick one; omit it and the system selects the default channel — normal calls are unaffected.

provider is not part of the official OpenAI protocol — it is a SeaWhale AI extension that only takes effect on this platform. The OpenAI SDK allows custom fields like this to pass through; see the examples below.

Value	Channel	Best for
direct	Direct	The official upstream link, for native behavior and the full context window
stable	Preferred	Balanced availability and speed — a good default for production traffic
economical	Economy	Cost first, well suited to batch processing and price-sensitive workloads

Available channels and their prices are listed under "Pricing" above (channels vary by model). Additional notes:

• Syntax: "provider": { "channel": "direct" }.
• If the requested channel is not enabled for that model, the request automatically falls back to the default channel and returns normally — it does not error.
• With the official SDKs: in Python pass it via extra_body; in Node.js put it directly on the request object and it passes through. In TypeScript projects, add a // @ts-expect-error line to skip the type check.

cURL

curl https://api.atalk-ai.com/v2/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer <API_KEY>" \
  -d '{
    "model": "doubao-seed-2-0-lite-260215",
    "messages": [
      {"role": "system", "content": "You are a helpful assistant."},
      {"role": "user", "content": "Hello!"}
    ],
    "provider": { "channel": "direct" },
    "stream": true
  }'
# provider is optional — remove this line to use the default channel

Python

from openai import OpenAI

client = OpenAI(
    base_url="https://api.atalk-ai.com/v2",
    api_key="<API_KEY>",
)

stream = client.chat.completions.create(
    model="doubao-seed-2-0-lite-260215",
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "Hello!"},
    ],
    stream=True,
    # Optional: pick a service channel; omit to use the default
    extra_body={"provider": {"channel": "direct"}},
)

for chunk in stream:
    if chunk.choices[0].delta.content:
        print(chunk.choices[0].delta.content, end="", flush=True)

Node.js

import OpenAI from 'openai'

const client = new OpenAI({
  baseURL: 'https://api.atalk-ai.com/v2',
  apiKey: '<API_KEY>',
})

const stream = await client.chat.completions.create({
  model: 'doubao-seed-2-0-lite-260215',
  messages: [
    { role: 'system', content: 'You are a helpful assistant.' },
    { role: 'user', content: 'Hello!' },
  ],
  stream: true,
  // Optional: pick a service channel; omit to use the default
  // @ts-expect-error provider is a SeaWhale AI extension, not in the OpenAI SDK types
  provider: { channel: 'direct' },
})

for await (const chunk of stream) {
  process.stdout.write(chunk.choices[0]?.delta?.content ?? '')
}