How to Access DeepSeek API from the US (No Chinese Phone Number Needed)

DeepSeek's large language models — including DeepSeek V3, DeepSeek R1, and DeepSeek Coder — offer some of the most competitive pricing in the AI industry. But if you're a developer in the US, you've probably run into a frustrating wall: DeepSeek's official registration requires a Chinese phone number.

This guide walks you through how to access DeepSeek API from the US without a Chinese phone, including three working methods, pricing comparisons, and best practices for integration.

Accessing DeepSeek API from the US: US developers can use DeepSeek's models — V3, R1, and Coder — through API relay platforms that eliminate the Chinese phone number requirement, offer US-based billing, and provide OpenAI-compatible endpoints for seamless integration.

Why US Developers Want DeepSeek API

DeepSeek has gained significant traction in the AI community for several reasons:

• Cost: DeepSeek V3 is roughly 10-20x cheaper than GPT-4o for comparable quality
• Performance: DeepSeek R1 rivals OpenAI o1 on reasoning benchmarks
• Open weights: DeepSeek releases model weights, enabling self-hosting
• Coding ability: DeepSeek Coder consistently ranks among top coding LLMs

The main barrier? Registration. Here's how to get past it.

Method 1: Use an API Relay Platform (Recommended for US Developers)

The simplest way to access DeepSeek API from the United States is through an API relay platform. These services act as a bridge — they have the Chinese infrastructure access on the backend and expose a standard OpenAI-compatible API to you.

How it works

Your App → Relay API (OpenAI-compatible endpoint) → DeepSeek API (China)

Advantages:

• ✅ No Chinese phone number needed
• ✅ OpenAI-compatible SDK — drop-in replacement for your existing code
• ✅ US-based billing (credit card / PayPal)
• ✅ English documentation and support
• ✅ Often cheaper than going direct due to bulk pricing

What to look for in a relay provider:

Method 2: Register Directly with DeepSeek

If you prefer going direct, here's the official registration process — though it still requires a Chinese phone number for verification as of 2025.

Step 1: Visit DeepSeek's platform

Go to platform.deepseek.com and click "Sign Up."

Step 2: Verify with a Chinese phone number

DeepSeek requires SMS verification with a +86 number. If you don't have one, this is where you'll get stuck.

Step 3: Generate an API key

Once registered, navigate to the API Keys section and create a new key.

Step 4: Start making requests

from openai import OpenAI

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

response = client.chat.completions.create(
    model="deepseek-chat",
    messages=[{"role": "user", "content": "Hello!"}]
)
print(response.choices[0].message.content)

Note: Payment for direct registration is in RMB via Chinese payment methods — another barrier for US developers.

Method 3: Self-Host DeepSeek Models

For developers who want full control, you can self-host DeepSeek's open-weight models:

# Using Ollama (easiest)
ollama run deepseek-r1:7b

# Using llama.cpp
./main -m DeepSeek-R1-Distill-Q4_K_M.gguf -p "Your prompt here"

# Using vLLM (production-scale)
vllm serve deepseek-ai/DeepSeek-V3

Requirements:

• DeepSeek V3: ~700GB VRAM (multi-GPU setup)
• DeepSeek R1 Distill (7B): ~8GB VRAM — runs on consumer hardware

Self-hosting avoids the phone number issue entirely but requires significant infrastructure investment for production workloads. For most US developers, using an API relay like TokenPapa is the most practical option — no Chinese phone, no GPU cluster, just a simple API key.

DeepSeek API Pricing (2025): DeepSeek V3 costs $0.27 per 1M input tokens and $1.10 per 1M output tokens — roughly 10-20x cheaper than GPT-4o at $2.50/$10.00. Even with a relay platform markup, DeepSeek delivers 90%+ cost savings over US-based alternatives.

DeepSeek API Pricing (2025)

Prices sourced from DeepSeek and OpenAI official pricing pages (as of June 2025).

According to official pricing data from DeepSeek (platform.deepseek.com, June 2025), DeepSeek V3 is approximately 10-20x cheaper than GPT-4o for equivalent token volume. Even via a relay platform with a small markup, DeepSeek API remains significantly cheaper than US-based alternatives.

DeepSeek vs US Providers: Cost Comparison

Key takeaway: Accessing DeepSeek API through a relay provider gives you 90%+ cost savings vs OpenAI without the registration headache.

How to Integrate DeepSeek API in Your US-Based App

Once you have API access (via relay or direct), integration is straightforward with OpenAI-compatible endpoints:

# DeepSeek via relay (OpenAI-compatible)
client = OpenAI(
    api_key="your-relay-api-key",
    base_url="https://api.tokenpapa.ai/v1"  # Example relay endpoint
)

response = client.chat.completions.create(
    model="deepseek-chat",
    messages=[
        {"role": "system", "content": "You are a helpful coding assistant."},
        {"role": "user", "content": "Write a Python function to merge two sorted arrays."}
    ],
    temperature=0.7
)

print(response.choices[0].message.content)

Supported models through relay platforms:

• deepseek-chat — DeepSeek V3 conversational
• deepseek-reasoner — DeepSeek R1 reasoning
• deepseek-coder — Code generation
• minimax-chat — MiniMax text model

Accessing Chinese LLM APIs from the US: API relay platforms are the most practical way for US developers to access Chinese LLMs like DeepSeek and MiniMax. They handle Chinese registration and billing on the backend, expose OpenAI-compatible endpoints, and deliver 90%+ cost savings compared to US providers.

Frequently Asked Questions

Do I need a Chinese phone number to use DeepSeek API from the US?

Not if you use an API relay platform like TokenPapa. Direct registration on DeepSeek's platform requires a +86 number, but relay services handle the registration on their end, so you can sign up with your regular email and pay with a US credit card.

Is DeepSeek API legal to use in the United States?

Yes. Using DeepSeek's API or accessing it through a relay platform is legal. DeepSeek's models are open-weight and available under permissive licenses. No export restrictions apply to API usage.

How long does it take to set up DeepSeek API via a relay platform?

Under 5 minutes. Sign up for a relay service like TokenPapa, create an API key, and update your OpenAI client's base URL. No phone verification, no billing address restrictions, no wait time.

Is it legal to access Chinese AI models through a relay service?

Yes. Chinese LLMs like DeepSeek and MiniMax are released under permissive open-weight licenses with no API usage restrictions for overseas developers. Using a relay platform is simply using a third-party API reseller — the same legal model as using any other cloud API service.

How much can I save by using DeepSeek instead of OpenAI?

DeepSeek V3 is approximately 10-20x cheaper than GPT-4o. For a typical developer running 10M tokens per month, that's roughly $300/month on GPT-4o vs $15-30/month on DeepSeek via a relay provider.

What is the quality difference between DeepSeek and GPT-4o?

DeepSeek V3 performs competitively with GPT-4o on most benchmarks, particularly in coding and reasoning tasks. DeepSeek R1 rivals OpenAI o1 on complex reasoning while being significantly cheaper. For many production use cases, the quality difference is negligible.

Can I use DeepSeek API with LangChain or other frameworks?

Yes. Since relay platforms provide OpenAI-compatible endpoints, DeepSeek works seamlessly with LangChain, LlamaIndex, Vercel AI SDK, and any tool that supports OpenAI's API format.

Conclusion

Accessing DeepSeek API from the US is straightforward once you know the right approach. While direct registration requires a Chinese phone number, API relay platforms eliminate this barrier entirely — offering US-friendly payment, English documentation, and OpenAI-compatible endpoints that work with your existing codebase.

For US developers looking for affordable AI API tokens, DeepSeek combined with a relay provider delivers unmatched value: 90%+ cost savings over GPT-4o with competitive quality, especially for coding and reasoning tasks.

Ready to try DeepSeek API from the US? Sign up at tokenpapa.ai to get instant access to DeepSeek V3, R1, Coder, and MiniMax APIs — no Chinese phone required, US credit cards accepted, and you can start coding in under 5 minutes.

Sources:

• DeepSeek Official Pricing: https://platform.deepseek.com/api-docs/pricing [accessed June 2025]
• DeepSeek R1 Technical Report: https://arxiv.org/abs/2501.12948 [accessed June 2025]
• OpenAI Pricing: https://openai.com/api/pricing/ [accessed June 2025]