Chinese models
DeepSeek, Kimi, Zhipu GLM, Qwen, MiniMax, Doubao, and models such as Meituan LongCat aggregated through OpenCode. They perform well in Chinese and are low priced, which suits Chinese conversation, coding assistance and batch processing.
Groups
| Group | Price |
|---|---|
deepseek | 100% of official price |
glm | 100% of official price |
kimi | 100% of official price |
minimax | 100% of official price |
opencode | 100% of official price |
qwen | Pay 50% of official price |
volcengine | 100% of official price |
Each vendor has its own group. Choose the token by the vendor you want to call.
Recommended API
Call them all through Chat Completions. The usage is the same as OpenAI, only model changes:
curl https://api.codingrouter.ai/v1/chat/completions \
-H "Authorization: Bearer $DEEPSEEK_GROUP_KEY" \
-H "Content-Type: application/json" \
-d '{"model": "deepseek-v4-flash", "messages": [{"role": "user", "content": "Write a quicksort in Python"}]}'DeepSeek
deepseek group. pro suits complex reasoning and coding; flash is fast and low cost.
Representative models: deepseek-v4-pro, deepseek-v4-flash.
Kimi
kimi group. kimi-for-coding is aimed at coding assistance; the highspeed versions output faster.
Representative models: kimi-k3.1-pro, kimi-k3, kimi-for-coding.
Zhipu GLM
glm group.
Representative models: glm-5.3, glm-5.3-flash. They output their thinking first, so leave enough max_tokens (2000 or more is recommended), otherwise they may return only thinking and an empty body.
Qwen
qwen group. max is the most capable; flash is fast and low cost.
Representative models: qwen3.8-max, qwen3.8-flash.
MiniMax
minimax group.
Representative model: MiniMax-M3.
Doubao
volcengine group, from Volcano Engine Ark.
Representative models: doubao-seed-2.1-pro, doubao-seed-2.0-mini.
OpenCode
opencode group, which aggregates models such as Meituan LongCat.
Representative models: longcat-2.0, muse-spark-1.3-contributor.
For the full list and prices see the model square.
