DeepSeek models: APIs and prices
3 models·Filter in the catalogue
DeepSeek models: open flagship quality at a low price. XiaoZhan AIhub offers 3 DeepSeek models (Text generation). Call them with one API key through the OpenAI, Anthropic or Gemini SDK; prices are in credits (60 credits = ¥1) and every model page lists the full price.
Text generation3
- DeepSeek V4 Flash
deepseek-v4-flashDeepSeek V4 Flash. 1M context, thinking on by default (reasoning_content stream). Billed at DeepSeek's official peak / off-peak prices: full price on weekdays 9:00-12:00 and 14:00-18:00 Beijing time, half price at all other times, including weekends and public holidays.
Input 0.12Credits/K·Output 0.48Credits/K - DeepSeek V4 Pro
deepseek-v4-proDeepSeek V4 Pro. 128K context, reasoning_content stream. Billed at DeepSeek's official peak / off-peak prices: full price on weekdays 9:00-12:00 and 14:00-18:00 Beijing time, half price at all other times, including weekends and public holidays.
Input 0.54Credits/K·Output 1.62Credits/K - DeepSeek V4.1 Flash
deepseek-v4.1-flashDeepSeek V4.1 Flash: a 552B-parameter MoE (asymmetric causal encoder-decoder) with native multimodal vision, 1M context and up to 384K output, thinking mode on by default, connected to DeepSeek directly. Billed at DeepSeek's official peak / off-peak prices: full price on weekdays 9:00-12:00 and 14:00-18:00 Beijing time, half price at all other times, including weekends and public holidays.
Input 0.12Credits/K·Output 0.48Credits/K
FAQ
- How do I call DeepSeek models?
- Create an API key on XiaoZhan AIhub, point your SDK's base_url at https://qd.zsopc.com/v1 and put the model ID from the list above in the model field. The OpenAI, Anthropic and Gemini SDKs all work.
- How are DeepSeek models billed?
- Each model is billed its own way: text models per input and output token, image models per image, video models per second or per token. Prices are in credits (60 credits = ¥1); each model page lists the full price.
Prices and details updated 2026-11-17.