ByteDance runs two product lines on the same Volcano Ark backbone: the Seedance video models and the Doubao Seed text family. With Seedance 2.5 making headlines, developers keep asking about the LLM side — here is the state of Doubao Seed 2 Pro and how to call it from outside China.
| Model | Best for | Context |
|---|---|---|
| doubao-seed-2.0-pro | Flagship reasoning, long-form, tool use | 256K |
| doubao-seed-2.0-lite | High-volume, cost-sensitive workloads | 256K |
| doubao-seed-2.0-code | Code generation & review | 256K |
All three are served official-direct through NovAI with USD billing — the upstream platform itself requires a Chinese business account and CNY payment, which is the main barrier for international developers.
from openai import OpenAI
client = OpenAI(api_key="sk-novai-...", base_url="https://aiapi-pro.com/v1")
r = client.chat.completions.create(
model="doubao-seed-2.0-pro",
messages=[{"role": "user",
"content": "Summarize this contract and list risk clauses."}],
max_tokens=4096,
)
print(r.choices[0].message.content)
Streaming, function calling and JSON mode all work through the same OpenAI-compatible surface. Live per-million pricing is on the marketplace page — it tracks the official rate card and is updated when ByteDance revises it.
The interesting property of the Doubao line is consistency: 256K context across Pro, Lite and Code means you can prototype on Pro, then drop the same prompts onto Lite for production volume without re-testing context behavior. For a cost-first alternative in the same ecosystem, ByteDance's own budget tier is worth benchmarking against DeepSeek and Qwen — see our cheapest-LLM roundup for the numbers.
Every new NovAI account starts with $2 free credit — enough to benchmark Seed 2 Pro against your current model on real prompts before committing.
Related: Seedance 2.5 video pricing · Cheapest LLM APIs this month