Understanding Our Pricing

DeepSeek v4-flash (deepseek-chat) uses a per-token pricing model:

• Input tokens (prompt + context): 1 CNY per million tokens
• Output tokens (model responses): 2 CNY per million tokens
• Cache-hit input tokens: 0.02 CNY per million (50x cheaper)

Cache hits occur when your input context overlaps with previously seen content. The longer your conversation history, the more cache hits you'll accumulate.

DeepSeek v4-flash supports a 1 million token (1M) context window — one of the largest available. Individual responses can be up to 384K output tokens. For most applications, outputs of 2K-8K tokens are typical.

Each AiCredits package represents a fixed amount of DeepSeek credits at DeepSeek's official rates, plus our service margin. This margin covers:

• Payment processing fees (card networks charge 2-3%)
• Operational overhead and server costs
• A small margin for sustainability

The result is still typically 90%+ cheaper than equivalent OpenAI API usage at current exchange rates.

Use the cost calculator on our homepage to estimate your monthly usage. As a rough guide:

• Light use (chatbots, small apps): $5-10/month
• Moderate use (content generation, analysis): $30-100/month
• Heavy use (data processing, agentic workflows): $100+/month

Start with a smaller package to estimate your actual consumption before committing to larger tiers.

DeepSeek may adjust their official rates at any time. AiCredits packages reflect current DeepSeek pricing at the time of purchase. If DeepSeek raises rates significantly, we will update our packages and notify existing customers. Package tokens are consumed at the rate in effect at the time of use.