What is the difference between DeepSeek-V3 and DeepSeek-R1?

DeepSeek-V3 is the fast general-purpose chat model; DeepSeek-R1 is the reasoning model that thinks step by step for harder problems and costs more, especially on output. Use V3 for most tasks and R1 only when you need deeper reasoning.

DeepSeek API — pricing, free tier & how to get a key

Q: Is the DeepSeek API cheap?

Yes — DeepSeek is among the cheapest capable LLM APIs. DeepSeek-V3 is roughly $0.27 per million input tokens and $1.10 output (reference, June 2026), far below GPT-4o, and DeepSeek also offers off-peak discounts and cached-input pricing that can cut costs further.

Q: How do I get a DeepSeek API key?

Create an account at platform.deepseek.com, open the API keys section, click create a new key, copy it once, and top up a small credit balance. The API is OpenAI-compatible, so most OpenAI SDKs work by changing the base URL.

DeepSeek's API is one of the cheapest ways to run a capable large-language model in production. It powers DeepSeek-V3 (fast chat) and DeepSeek-R1 (step-by-step reasoning), is OpenAI-compatible, and undercuts most US providers on price. Here's what it costs and how to get your key.

DeepSeek API pricing (reference, June 2026)

Model	Input $/1M	Output $/1M	Best for
DeepSeek-V3 (chat) cheapest	$0.27	$1.10	General tasks, high volume
DeepSeek-R1 (reasoner)	$0.55	$2.19	Hard reasoning, math, code
V3 — cached input	$0.07	—	Repeated context (big discount)

⚠️ Reference prices, June 2026 — DeepSeek updates pricing and runs off-peak discounts. Confirm on api-docs.deepseek.com before budgeting. Prices are per 1M tokens; output is billed separately and costs more. Cached-input (repeated context) is billed at a large discount.

→ Estimate your bill on the AI API cost calculator (DeepSeek-V3 is included) or model a whole app with the AI app cost estimator.

Is there a free tier?

DeepSeek has no permanent free tier, but pricing is so low that a small top-up lasts a long time, and new accounts sometimes get a modest starting credit. If you specifically need a free quota with no card, Google Gemini is the better fit; if you just want the lowest pay-as-you-go cost, DeepSeek-V3 is hard to beat.

How to get a DeepSeek API key (step by step)

1. Go to platform.deepseek.com and create an account.
2. Open API keys and click Create new API key; copy it once (you can't view it again).
3. Add a small credit top-up under Billing — usage is deducted from your balance.
4. Point your code at the DeepSeek base URL. It's OpenAI-compatible, so existing OpenAI SDKs work by swapping the base URL and key.

Test it with a simple request:

# quick test (replace $DEEPSEEK_API_KEY)
curl https://api.deepseek.com/chat/completions \
  -H "Authorization: Bearer $DEEPSEEK_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"deepseek-chat","messages":[{"role":"user","content":"hi"}]}'

Estimate your cost

Use the AI API Cost Calculator to plug in your token counts and request volume — it ranks DeepSeek against GPT-4o, Claude and Gemini from cheapest to most expensive for your exact workload.

Cheaper alternatives

DeepSeek is already at the cheap end. The closest rivals are Gemini 2.0 Flash ($0.10/$0.40) and GPT-4o mini ($0.15/$0.60) for simple high-volume tasks. For reasoning on a budget, DeepSeek-R1 competes with much pricier models like o3.

FAQ

Is the DeepSeek API cheap?

Yes — among the cheapest capable APIs. DeepSeek-V3 is ~$0.27/$1.10 per 1M tokens, with off-peak and cached-input discounts that cut costs further.

How do I get a DeepSeek API key?

Sign up at platform.deepseek.com, create a key in the API keys section, copy it once, and add a small credit balance. The API is OpenAI-compatible.

DeepSeek-V3 vs R1 — which should I use?

V3 for most chat and generation tasks; R1 (the reasoner) for hard math, logic and code where step-by-step thinking helps — it costs more, especially on output.

Can I use OpenAI libraries with DeepSeek?

Yes. DeepSeek's API is OpenAI-compatible — change the base URL to api.deepseek.com and use your DeepSeek key.

Not affiliated with DeepSeek. Prices are reference estimates — always verify on the official pricing page.