"Free" With DeepSeek: What's Real
DeepSeek can be inexpensive for some workloads, but “free chat,” promotional API credit, and self-hosted weights are different offers. Here is how to separate them before sending real data or forecasting production cost.
The Free & Nearly-Free Routes
- Consumer chat. The web and mobile products may offer no-cost interactive usage subject to current limits. They are not a production API.
- Open weights (free to download, not free to operate). DeepSeek V4 is MIT-licensed. Self-hosting replaces per-token API billing with accelerator capacity, storage, deployment, monitoring, and engineering cost. Check the official model card and inference project for the memory required by the exact checkpoint and precision you choose.
- Trial credits. New API accounts and gateways typically grant starter credit. LinkModel gives a $1 signup credit across models including
deepseek-v4-flash— enough for a lot of Flash tokens given its pricing. - Prefix caching. Repeated stable prefixes may receive a lower cached-input rate. Confirm the current rate in DeepSeek API pricing, then place stable instructions before variable user content.
Why "Nearly Free" Is the Real Story
DeepSeek can be inexpensive compared with premium models, but its direct API uses peak and off-peak time bands. Current published rates per 1M tokens are:
| Model | Input | Output | Cached input |
|---|---|---|---|
| DeepSeek V4 Flash | $0.22 off-peak / $0.44 peak | $0.66 off-peak / $1.32 peak | $0.007 off-peak / $0.014 peak |
| DeepSeek V4 Pro | $0.66 off-peak / $1.32 peak | $1.98 off-peak / $3.96 peak | $0.022 off-peak / $0.044 peak |
Both models expose 1M-token context, and the cached-input tier can materially reduce repeated-prefix cost. Compare the active time band and your measured cache-hit ratio rather than quoting one promotional number. Full comparison in DeepSeek V4 Flash vs GPT-4o.
The Caveat: Data Residency
For EU/PII or regulated workloads, verify the DeepSeek API's current processing location and privacy terms. Self-hosting changes the infrastructure boundary; a gateway adds its own data-handling terms. Do not assume either route is compliant without reviewing the actual deployment and contract. Confirm active model IDs in the provider or gateway documentation instead of relying on legacy aliases in an older article.
Your First Call
DeepSeek's API is OpenAI-compatible, so it's a base-URL + model-name change from an OpenAI setup. Through a gateway you get one key across chat models:
curl -X POST https://api.linkmodel.ai/v1/chat/completions \
-H "Authorization: Bearer $LINKMODEL_API_KEY" \
-H "Content-Type: application/json" \
-d '{ "model": "deepseek-v4-flash", "messages": [{"role":"user","content":"Summarize this in 3 bullets: ..."}] }'Confirm the exact chat endpoint and request schema in the docs and model catalog before wiring it up.
Free Chat Is Not a Free Production API
Separate the DeepSeek consumer chat interface, promotional API credits, third-party trials, and self-hosted weights. Each has different limits and terms. A browser chat plan cannot be automated as an API, and a one-time credit is not a durable production budget.
Before using any “free key” offer, verify the provider, expiration date, card requirement, rate limit, model version, data handling, and what happens when the credit ends. Never paste a production secret into an unofficial key-sharing site.
A Safe Trial Plan
Use trial credit to answer four questions:
- Does the documented model ID work for the account and region?
- Do structured outputs and tool calls pass your application tests?
- What is the measured cost per successful task after retries?
- Does the provider's data policy fit the workload?
Run 20–50 representative requests, log tokens and failures, then forecast paid usage. A nearly free model can be a good default, but only when it consistently completes the task.
Protect the Key
Keep the API key in a server-side environment variable, add spending alerts, and apply per-user rate limits at your backend. Rotate the key if it appears in a repository or client bundle. Free credit attracts abuse just as quickly as paid balance.
Bottom Line
- Just chatting/testing? The free consumer app or a $1 gateway credit.
- Want truly free at scale? Self-host the open weights.
- Building cheaply? Measure the current input, cached-input, output, and retry cost per successful task; do not treat a promotional or cached rate as permanently free.
For where DeepSeek sits against everyone else, see the AI API pricing comparison and cheapest AI API guide.
Start free with a $1 credit and run V4 Flash today.
This is factual pricing/how-to information; verify current rates and data-handling terms on each provider's official pages before production use.