How to get a DeepSeek API key
Updated · 2 min read
DeepSeek offers capable models at very low per-token prices, and its API speaks the same formats as OpenAI and Anthropic. That makes a DeepSeek key one of the cheapest ways to power BYOK apps, and it works in many apps even without a dedicated DeepSeek option.
Step by step
- Create an account on the DeepSeek platform and open platform.deepseek.com/api_keys.
- Create an API key, name it after the app, and copy it.
- Top up your balance. Usage is deducted from your topped-up balance (and from any granted balance first).
- Add the key to your app. If the app lists DeepSeek, pick it. Otherwise use a custom OpenAI-compatible provider with base URL
https://api.deepseek.com(or the Anthropic-compatiblehttps://api.deepseek.com/anthropic) and the model namedeepseek-flashordeepseek-v4-pro.
What it costs
DeepSeek charges per million tokens and halves its prices outside peak hours. Peak hours are 01:00–04:00 and 06:00–10:00 UTC, Monday to Friday (excluding Chinese public holidays); all other hours, including weekends, are off-peak. Prices on 2026-10-23:
| Per 1M tokens | deepseek-flash (peak / off-peak) | deepseek-v4-pro (peak / off-peak) |
|---|---|---|
| Input, cache hit | $0.006 / $0.003 | $0.044 / $0.022 |
| Input, cache miss | $0.30 / $0.15 | $1.32 / $0.66 |
| Output | $1.20 / $0.60 | $3.96 / $1.98 |
A chat turn of about 2,000 tokens in and 500 out costs roughly $0.0012 with deepseek-flash at peak, so $5 covers about 4,000 turns. Repeated context (cache hits) is billed at a small fraction of the normal input price. Both models accept up to 1M tokens of context; Flash also handles images.
Things to know
- Off-peak is half price. If an app runs long jobs, scheduling them outside peak hours halves the bill.
- Data location. DeepSeek is a Chinese company; check its privacy policy before sending confidential material. Gateways such as OpenRouter can serve DeepSeek models from other hosts.
- Limits. DeepSeek publishes concurrency limits per model (2,500 concurrent requests for Flash, 500 for Pro on 2026-10-23), which are far above what a personal app needs.
- Keep keys separate. One key per app, revoke unused keys, and never paste a key into public code.
Apps that work with your DeepSeek API key
Cheaper still
For zero-cost options (Gemini's free tier, OpenRouter free models, local models), see free AI API keys. For running open models on your own computer, see local LLM apps.
Questions
Is the DeepSeek API free?
No. You top up a balance and usage is deducted from it; any granted (promotional) balance is used first. Prices are among the lowest of the major providers, and off-peak rates are half price.
Which model name should I use?
deepseek-flash for the fast, cheap model (DeepSeek-V4.1-Flash) and deepseek-v4-pro for the larger one. The old names deepseek-v4-flash and deepseek-v4-flash-vision-exp still work and are billed at the Flash price.
My app has no DeepSeek option. Can I still use it?
Often yes. DeepSeek's API is compatible with the OpenAI format (base URL https://api.deepseek.com) and the Anthropic format (https://api.deepseek.com/anthropic), so apps that accept a custom OpenAI-compatible or Anthropic-compatible endpoint can use it.
Where is my data processed?
DeepSeek is a Chinese company. Read its privacy policy before sending anything sensitive, or reach DeepSeek models through a gateway such as OpenRouter that routes to other hosts.
Sources
Checked against the providers' own documentation on 2026-10-23. Prices and limits change; the links below have the current figures.