Chinese model APIs on one OpenAI-compatible key. Browse models

Pricing landing · last reviewed 2026-11-09

DeepSeek API pricing in USD

Overseas developers searching for DeepSeek API pricing usually need three facts: the model ID, the USD rate per 1M tokens, and a Base URL that works without Chinese phone signup. SmarToken publishes Live DeepSeek routes with prepaid USD billing, cache-aware rates and OpenAI-compatible requests.

DeepSeek reference prices use peak hours and uncached input. Off-peak rates are 50% lower; cached input is billed at its discounted rate. View all DeepSeek rates

Model IDNameStatusInput / 1MCache hit / 1MOutput / 1M
deepseek-chatDeepSeek FlashLive$0.3$0.006$1.2
deepseek-v4-proDeepSeek V4 ProLive$1.32$0.044$3.96

Displayed USD rates have no additional platform surcharge. Peak Beijing weekday windows can raise DeepSeek upstream tariffs; off-peak can be lower. Exact billed usage is recorded after each response or stream finishes.

Facts for buyers

  • Wallet denomination: prepaid USD
  • OpenAI-compatible Base URL: https://thesmartoken.com/v1
  • Live DeepSeek IDs: deepseek-chat, deepseek-v4-pro
  • No Chinese phone required for Google signup
  • Key-level daily and monthly budgets available

What changes cost

  • Prompt length and retrieved context (input tokens)
  • Answer verbosity and reasoning traces (output tokens)
  • Cache hits on repeated system prompts
  • Peak vs off-peak Beijing time windows
  • Retries after client errors — stop on HTTP 402

Minimal priced call

curl --fail-with-body -sS "https://thesmartoken.com/v1/chat/completions" \
  -H "Authorization: Bearer $SMTOKEN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek-chat",
    "messages": [{"role":"user","content":"Estimate this request cost."}]
  }'

After the call, inspect token counts and USD debit in the console. If the wallet is empty, the API returns HTTP 402 — stop automatic retries and top up first.

FAQ

What is DeepSeek API pricing on SmarToken?
SmarToken bills DeepSeek routes in prepaid USD per 1M tokens. deepseek-chat (DeepSeek Flash) and deepseek-v4-pro publish separate input, cached-input and output rates. Peak Beijing weekday windows can be higher than off-peak.
Is DeepSeek API pricing charged in CNY?
No on SmarToken. The customer wallet and usage logs are USD. You do not need a Chinese bank account to top up or inspect spend.
Which model IDs should I price against?
Use deepseek-chat for DeepSeek Flash and deepseek-v4-pro for DeepSeek V4 Pro. Confirm live availability with GET /v1/models before production traffic.
How do I keep DeepSeek API cost predictable?
Top up the prepaid USD wallet, set daily or monthly API key budgets, watch token counts in the console, and compare DeepSeek with Kimi, Qwen or GLM when quality allows a cheaper route.
Where is the longer budget guide?
See /topics/deepseek-api-pricing for token-budget planning narratives, and /pricing for the full gateway price table.