Multimodal Chinese AI APIs

As of 2026-11-04.

• MiniMax — text plus speech/video products on minimax.io pricing.
• StepFun — vision token pricing and extensive speech/music SKUs on platform.stepfun.ai (note StepAudio free-window end date on official docs).
• DeepSeek — Flash supports vision per API docs; Pro does not.
• Qwen — VL and other multimodal SKUs via Model Studio catalogs.

Audio/video billing units (minutes, characters, clips) are not comparable to text tokens without converting to your workload.