If you have landed here, you probably want one thing: a working DeepSeek account, ideally in the next few minutes, without a hidden paywall at the end. This guide walks you through the DeepSeek sign up process for the web chat, the iOS and Android apps, and the developer platform that issues API keys. I have registered accounts on each surface using UK and US phone numbers and email, and I will flag the bits that trip people up — regional availability, verification codes that never arrive, and the difference between a chat account and a billing-enabled API account. By the end you will know exactly which route to pick, what it costs, and what to do first once you are in.

Where to sign up, in one sentence

For the chatbot, register at chat.deepseek.com or inside the official mobile app; for API access, register separately at platform.deepseek.com. The two accounts share credentials but serve different purposes — the chat surface is free to use with session history, and the developer platform requires a billing top-up before your API key will return a response.

Which DeepSeek account do you actually need?

Before you hand over an email address, work out which surface fits your use case. Most readers need only the first row.

The web and app accounts maintain your conversation history server-side. The API is a separate, stateless surface — your client resends the full message history on every call, which I will come back to in the developer section. If you are new to the ecosystem, the DeepSeek beginners guide covers the wider map before you pick a route.

Sign up for the DeepSeek web chat

The web flow takes under two minutes on a decent connection. You will need an email address or a phone number that can receive SMS.

1. Open chat.deepseek.com in any modern browser. Chromium-based browsers and Safari both work; I tested Firefox 126 and Edge 125 without issues.
2. Click “Sign up” in the top-right. The toggle switches between Email and Mobile registration.
3. Enter your email or phone number, choose a password of at least eight characters (mix case and numbers — the validator is strict), and tick the terms box.
4. Enter the verification code that arrives by email or SMS. Codes expire after about 10 minutes; if none arrives, check spam, then request a new one after 60 seconds.
5. Set a display name when prompted. You are now signed in and the current default model is DeepSeek V4.

You can also register with a Google account on the web, which skips the verification step. Apple Sign In is available in the mobile app. Once you are in, the DeepSeek chat walkthrough explains the DeepThink toggle, file upload, and search features you will see in the sidebar.

What does the free web account include?

The web chat is free to use and defaults to V4. There is a DeepThink toggle that switches V4 between non-thinking and thinking mode — thinking mode returns a reasoning trace and is slower but stronger on multi-step problems. DeepSeek does not publicly document a fixed daily message cap as of April 2026, but expect soft rate limits during peak hours. For a breakdown of what falls inside and outside the free tier, the is DeepSeek free guide has the current picture.

Sign up on the mobile app

The iOS and Android apps share the same account system as the web. If you already registered on the web, you can skip creating a new account and just log in.

• iPhone: open the App Store, search “DeepSeek — AI Assistant”, confirm the publisher is DeepSeek, and install. The registration flow mirrors the web.
• Android: open Google Play, search “DeepSeek”, and confirm the developer before installing. Outside Google Play, only use APKs from the official site — see verify official DeepSeek app to avoid the copycat apps that flood third-party stores.
• Sign-in methods: email, phone, Google (Android), and Apple Sign In (iOS). Phone-number sign-in works with most international carriers; I registered a +44 UK number and a +1 US number without issue.

If the app refuses to install or shows “not available in your region”, consult DeepSeek availability by country. Italy, for example, has had restrictions in place since January 2025, and several US states have restricted the app on government devices.

Sign up for the DeepSeek API

The developer platform is a separate account system at platform.deepseek.com. Even if you already have a chat account, you will register again here, verify your details, and add payment before an API key will work against requests.

1. Go to platform.deepseek.com and click “Sign up”.
2. Register with email or phone. Phone-based accounts in some regions are required for billing, depending on your card issuer.
3. Verify via the code sent to your email or phone.
4. Open the “API Keys” section in the left sidebar and click “Create new secret key”. Copy it immediately — DeepSeek shows the key once.
5. Add a billing top-up under “Billing”. A small top-up (for example $5) is enough to test at V4-Flash rates for a long time.

DeepSeek may offer a granted balance — a small promotional credit that can expire; check the billing console for current offers before assuming you have free quota. The step-by-step in get a DeepSeek API key shows the console screens, and DeepSeek API authentication covers header format and key rotation.

Your first API call

The DeepSeek API is OpenAI-compatible, so the official OpenAI SDK works by swapping base_url and api_key. Chat requests hit POST /chat/completions, the OpenAI-compatible endpoint. DeepSeek also ships an Anthropic-compatible surface against the same base URL, which was added alongside V4 on April 24, 2026.

Here is a minimal Python example against deepseek-v4-flash:

from openai import OpenAI

client = OpenAI(
    base_url="https://api.deepseek.com",
    api_key="YOUR_KEY_HERE",
)

resp = client.chat.completions.create(
    model="deepseek-v4-flash",
    messages=[{"role": "user", "content": "Say hello in three languages."}],
)
print(resp.choices[0].message.content)

Two things to know from day one. First, the API is stateless — you must resend the full conversation history with every request. That is different from the web app, which keeps your history server-side. Second, the current generation is DeepSeek V4, released April 24, 2026, and shipped as two model IDs: deepseek-v4-pro (1.6T total / 49B active parameters, frontier tier) and deepseek-v4-flash (284B / 13B active, cost-efficient tier). Both are open-weight Mixture-of-Experts under the MIT license, with a 1,000,000-token context window and up to 384,000 tokens of output.

Legacy IDs and the migration window

If you follow older tutorials that reference deepseek-chat or deepseek-reasoner, those IDs still work — they currently route to deepseek-v4-flash in non-thinking and thinking mode respectively. They retire on 2026-07-24 at 15:59 UTC, after which requests using those IDs will fail. Migration is a one-line change: update the model= field; base_url does not change.

Thinking mode is a parameter, not a model

In V4, thinking mode is toggled per request. The same model ID handles both fast and reasoning-heavy work:

• Non-thinking (default): no reasoning_effort field; fastest and cheapest per request.
• Thinking (high): reasoning_effort="high" with extra_body={"thinking": {"type": "enabled"}}.
• Thinking-max: reasoning_effort="max" for the deepest reasoning effort.

When thinking is enabled, the response returns reasoning_content alongside the final content. Other parameters worth knowing on day one include temperature (DeepSeek recommends 0.0 for code and math, 1.3 for general conversation, 1.5 for creative writing), top_p, max_tokens, JSON mode, tool calling, streaming, and context caching. JSON mode is designed to return valid JSON — include the word “json” and an example schema in your prompt, set max_tokens high enough to avoid truncation, and handle the occasional empty content response.

Pricing snapshot: what sign-up actually costs

The web chat and apps are free. The API is pay-as-you-go with no flat subscription. Rates, as listed on DeepSeek’s pricing page at the time of writing (April 2026):

Off-peak discounts ended on 2025-09-05 and did not return with V4. For a concrete worked example on deepseek-v4-flash: 1,000,000 calls with a 2,000-token cached system prompt, a 200-token user message, and a 300-token response cost $56.00 (cached input) + $28.00 (uncached input, each new user message is a miss) + $84.00 (output) = $168.00 total. The same workload on deepseek-v4-pro lands at $1,682.00. Always count all three token buckets when estimating spend. The DeepSeek API pricing breakdown has the full rate card.

Troubleshooting sign-up problems

The issues readers report most often, in roughly decreasing frequency:

• No verification code: wait 60 seconds and request again. SMS delivery can stall on some mobile carriers — switching to email usually works.
• “Email already registered”: you have a chat account and are trying to create a duplicate on the same platform. Use DeepSeek login instead, or reset DeepSeek password if you have forgotten it.
• “Service not available in your region”: the Italian block from January 2025 remains enforced for the consumer app, and several US states restrict government use. Check the availability by country page before fighting the error.
• API key returns 401 or 402: either the key is wrong or your balance is zero. Open the billing page and top up.
• Phone number rejected: some virtual numbers (Google Voice, second-line apps) are blocked. Use a native carrier number.

If the app itself misbehaves after sign-up, DeepSeek troubleshooting collects the fixes I have needed to apply in the last year.

What to do in your first 15 minutes

Once the account is live, do these five things before anything else. They are the difference between “I signed up and forgot about it” and actually extracting value.

1. Run a DeepThink query. Toggle thinking mode on and ask a multi-step question — something your usual chatbot gets wrong. This is the fastest way to feel the difference between V4-Flash default and V4-Pro with thinking enabled.
2. Upload a file. Try a PDF or a spreadsheet. The 1M-token context means “upload the whole thing” is rarely a problem.
3. If you have an API key, run a curl call. Confirm the end-to-end path works before you build anything on top.
4. Set a password manager entry for both the chat and platform accounts. They are separate logins.
5. Read the privacy terms. Your conversations are processed on servers subject to Chinese law; the DeepSeek privacy guide summarises the trade-offs honestly.

For a wider onboarding path after you are set up, see the rest of our DeepSeek beginner guides.

Last verified: 2026-04-24. DeepSeek AI Guide is an independent resource and is not affiliated with DeepSeek or its parent company. Model IDs, pricing and API behaviour change; check the official DeepSeek documentation and pricing page before committing to a production decision.