Cheapest Chinese AI models by category

The lowest list price in each category across 111 Chinese models, at provider list price with a 0% platform fee. 5 models are free outright.

Cheapest Chat / LLM (top 5 by list price)

ModelProviderList priceUnit
Hunyuan MT2 LiteTencent$0.043flat
Hunyuan MT2 PlusTencent$0.072flat
Doubao-Seed-2.0-LiteByteDance$0.075flat
GLM-5.3-FlashZhipu AI$0.080flat
GLM-4.6V APIZhipu AI$0.100per 1M tokens

Cheapest Image (top 5 by list price)

ModelProviderList priceUnit
Wan 2.2 T2I FlashAlibaba Cloud$0.021per generation
WAND Vega Image LiteTencent (WAND Vega)$0.025per generation
Seedream 5.0ByteDance$0.027per generation
Vidu Image Q2Vidu (Shengshu)$0.029per generation
Wan 2.5 T2I PreviewAlibaba Cloud$0.031per generation

Cheapest Video (top 5 by list price)

ModelProviderList priceUnit
Seedance 2.0 FastByteDance$0.054flat
Seedance 2.0ByteDance$0.067per generation
Seedance 2.5ByteDance$0.140per generation
Qwen3 Omni FlashAlibaba Cloud$0.150per image
Hunyuan Turbos Vision VideoTencent$0.462per image

Cheapest 3D (top 5 by list price)

ModelProviderList priceUnit
Hunyuan 3D MotionTencent$0.185per generation
Hunyuan 3D ExpressTencent$0.277per generation
Tripo 3D P1Tripo$0.323per generation
Tripo 3D 3.1Tripo$0.323per generation
Hunyuan 3D 3.1Tencent$0.369per generation

Cheapest Audio (top 5 by list price)

ModelProviderList priceUnit
Qwen3 TTS FlashAlibaba Cloud$0.015flat
MiniMax Speech 2.6 HDMiniMax$1.54flat
Hunyuan ASR 3.0Tencent (Hunyuan)$1.54flat
MiniMax Speech 02 HDMiniMax$1.54flat
MiniMax Speech 02 TurboMiniMax$1.54per generation

Cheapest Embedding (top 5 by list price)

ModelProviderList priceUnit
Kinfra VL Embedding 2BTencent (Kinfra)$0.077per image
Kinfra Text Embedding 0.6BTencent (Kinfra)$0.077flat
Text Embedding V4Alibaba Cloud$0.080flat
Kinfra Text Embedding 4BTencent (Kinfra)$0.092flat
Kinfra VL Embedding 8BTencent (Kinfra)$0.092per image

Questions

What is the cheapest Chinese AI API?

The cheapest paid models start at $0.015 per unit, and 5 of the 111 models are free - including GLM-4.7-Flash for text and CogVideoX-Flash for video (5 generations a day, no card).

Cheapest is not always cheapest per finished result

A cheaper per-call model that needs more retries can cost more than a pricier one that lands first time. The cost tables include the retry multiplier for that reason.

Do I need a different key per model?

No. All of these sit behind one OpenAI-compatible endpoint and one API key.

Try the cheapest models free first
Five free models, 10 calls a day, no card. One key for all 111.
Get a free key Full price table →

Generated 2026-11-01 19:00 UTC from the machine-readable price feed. Prices belong to their respective providers; markup is 0% and checkable at transparency.