Updated
DeepSeek API Key: Free or Paid? Real Costs
People asking whether a deepseek api key is free or paid are usually mixing up two things: getting a key and using it. Creating a key is normally free on API platforms. Sending requests costs money once any free credit is gone.
Key versus usage
An API key identifies you. It does not carry a price. The cost sits in the billing model behind it, and that model usually has three parts: any free or promotional credit, a prepaid balance or card on file, and a per-token rate for input and output. For the current numbers at DeepSeek or any other provider, check their current pricing page, since rates and promotions change.
So the honest answer to the question is: the key is typically free to create, heavy use is paid, and small tests may be covered by starter credit where a provider offers it. Whether such credit exists, how large it is and when it expires is the provider's decision, and you should read it from their dashboard rather than from a blog post, including this one.
Free chat is not a free API
Many people hear "free" from a chat website and expect the same from the API. These are separate products. A web chat gives a human a text box with limits the site chooses. An API gives a program direct access to the model, counts every token, and bills it.
| Free web chat | API key | |
|---|---|---|
| Who uses it | A person in a browser | Code, bots, apps, editor plugins |
| Billing | Usually none, with usage caps | Per token, from balance or crypto |
| Control | Fixed interface and rules | Your own prompt, parameters, history |
| Automation | Not intended | The whole point |
If you only need a few answers a day for yourself, a free chat may be all you need. If you are wiring a model into SillyTavern, Janitor AI, a script or an editor, you need an API and a balance.
How per-token billing works
A token is a chunk of text, roughly three-quarters of an English word on average. You pay for tokens in the prompt (input) and tokens in the reply (output) at separate rates. The history you resend every turn counts as input each time, which is why long chats cost more per message than short ones.
The formula is simple: cost = input_tokens / 1,000,000 x input_rate + output_tokens / 1,000,000 x output_rate. Once you know the rates, any workload can be priced on paper.
What it costs on our endpoint
This site sells access to a different model from DeepSeek's. It is one open-weight model run by an independent provider and tuned to answer without content refusals for lawful adult use. It is not DeepSeek's official API. The rates are $0.15 per 1M input tokens and $0.50 per 1M output tokens. Worked examples:
- One short question: 500 input tokens and 400 output tokens. 500 / 1,000,000 x $0.15 = $0.000075. 400 / 1,000,000 x $0.50 = $0.0002. Total $0.000275.
- A chat of 100 turns, each with a 4,000-token prompt and a 300-token reply: input 400,000 tokens = $0.06; output 30,000 tokens = $0.015; total $0.075.
- A batch job of 10,000 documents at 1,000 input and 200 output tokens each: input 10,000,000 tokens = $1.50; output 2,000,000 tokens = $1.00; total $2.50.
Failed or refused requests are not charged. Credit is prepaid, there is no subscription, and it never expires.
Getting started without paying
New users can claim $0.50 of trial credit with one click on the API key page after sign-up with Google or email. It needs no card, lasts 7 days and is limited to one per person. At the rates above, $0.50 covers about six of the 100-turn chats, or roughly 1,800 of the short questions. The trial key has lower limits than a full key: 2 concurrent requests and 60 requests per minute.
curl https://api.deepseekapikey.com/v1/chat/completions \
-H "Authorization: Bearer YOUR_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"uncensored","messages":[{"role":"user","content":"Say hello in five words."}]}'If the call returns a reply, your key, URL and model all work. If it does not, the documentation lists the status codes and limits, and the support page can help with the rest.
Track your own spend
Do not wait for a surprise on the balance page. Every non-streamed response includes a usage object with prompt and completion tokens, and streamed responses report usage at the end. Multiply by the rates and log it.
from openai import OpenAI
client = OpenAI(base_url="https://api.deepseekapikey.com/v1", api_key="YOUR_KEY")
r = client.chat.completions.create(
model="uncensored",
messages=[{"role": "user", "content": "Summarize TCP in two sentences."}],
)
u = r.usage
cost = u.prompt_tokens / 1_000_000 * 0.15 + u.completion_tokens / 1_000_000 * 0.50
print(u.prompt_tokens, u.completion_tokens, round(cost, 6))Add that to a log line per request and you have a cost report by the end of the week. It also catches the most common budget leak: a loop that resends a growing history without trimming it.
Topping up when the trial ends
Top-ups are any whole amount from $20 to $500. Bonuses apply: +5% from $50, +10% from $100, +15% from $200. For example, a $100 top-up gives you $110 of credit, and a $200 top-up gives you $230. Payment is by crypto (USDT on TRC20, USDC on Base).
At the 100-turn chat example costing $0.075, a $20 top-up pays for about 266 such chats. Your own usage will differ, so run a day of real traffic on the trial or a small top-up and read the X-Balance-USD header to see actual spend before you commit more.
Choosing between free, cheap and premium
Decide by workload. For occasional personal use, a free chat site is fine. For tools that need an OpenAI-compatible endpoint, any prepaid per-token API works, and your task is to compare what each charges and what each refuses. Look at the output rate in particular, because long replies are priced higher than long prompts. And read the provider's content policy: if your application involves adult fiction, a model that refuses it will cost you time even when it is cheap.
The full table of rates and bonuses is on the pricing page. For setup in tools, see the SillyTavern guide.
More guides
Questions and answers
Is it free to create a DeepSeek API key?
Creating a key on API platforms is usually free, but check DeepSeek's own dashboard and pricing page for current terms. Using it is billed per token.
Does our endpoint charge a monthly fee?
No. It is pay as you go from prepaid credit with no subscription, and the credit never expires.
Are failed requests billed?
No. Failed or refused requests are not charged on our endpoint.
Is there a free trial?
Yes. New users can claim $0.50 of trial credit with one click on the API key page. It lasts 7 days and needs no card.
Your key is one form away
Create an account, copy the key, change the base URL. That is the whole setup.
