Pricing
Pay-as-you-go, metered per call. Currency: CNY per million tokens. Balance never expires.
Input tokens
¥2.00
Per million tokens · the prompts and context you send to the model
Output tokens
¥8.00
Per million tokens · text generated by the model (output priced at 4.0× input)
Cache hit (read)
¥0.10
Vendor list cache price · keep the system prompt stable to hit the cache
Cache hits cost less: ¥0.10 per million (about 1/20 of the input price) — save about 95% on repeated prefixes
Prices are as published on this site.
Live pricing follows GET /v1/models responses; the console itemizes the input, output and cached tokens of every call.
Capabilities & use cases
Overview
The DeepSeek V4 family (1.6T total / 49B active) matches top closed-source models in performance, with a native 1M context, and stands out at code generation and long-horizon reasoning. DeepSeek-V4-Flash (284B / 13B active) delivers high efficiency at a lower price.
Key strengths
Code generation & refactoringLong-horizon math reasoningNative 1M contextCache hits save 95%
Recommended use cases
Everyday chat, batch text processing, and cost-sensitive workloads that still need consistent quality.
Example call
Fully OpenAI-compatible — just point base_url at ShinLinkAI. No changes to your application code.
curl
curl https://api.shinloo.com/v1/chat/completions \
-H "Authorization: Bearer sk-sl-…" \
-H "Content-Type: application/json" \
-d '{
"model": "deepseek-v4-flash",
"messages": [{"role": "user", "content": "Hello"}]
}'