Grok API from Russia: connecting without a foreign card

If you need the Grok API from Russia, you are probably stuck on payment: direct access to the xAI API has its own registration and payment terms, and they may not work for you (check the current ones on the xAI website). Through the Zukko gateway, Grok models are available in the same OpenAI-compatible format as the rest, and you can pay by SBP, MasterCard or crypto. Below is the whole path from registration to the first answer, with code and a price calculation.

Which Grok models are available

Zukko prices as of October 18, 2026, dollars per 1M tokens. "Strength" is the Intelligence Index (Artificial Analysis).

Grok has a 256K-token window, smaller than the 1M of most catalogue models; for very long documents this matters. As for cache, reads are counted if the supplier reports them in the response; there is no separate charge for cache writes. The current prices and list are always at GET https://api.zukko.pro/v1/models.

Step 1. Balance

Register at zukko.pro↗ and top up your balance. Access from Russia goes through the gateway; top-ups start from $1:

• SBP: the amount is charged in rubles at the market rate at the time of payment, the payment page adds an 8.5% fee on top, and the dollar amount you specified is credited to the balance. For example, to get $5 on your balance you pay 5 × 1.085 = $5.425 in rouble equivalent. Step-by-step guide: paying through SBP↗.
• Cryptocurrency (USDT and others).
• MasterCard: a fee of about 13%, amounts from $2 to $594.

There is no subscription: you pay for the tokens you use.

Step 2. Key

In the API keys↗ section create a key. The full fetch_... key is shown once, so save it. Set a daily limit in dollars right away and, if you like, allow the key only Grok models: this guards against accidental spend.

Step 3. Address

For OpenAI-compatible clients the base URL is https://api.zukko.pro/v1. Authorization is the header Authorization: Bearer fetch_.... Clients that add /v1 themselves follow a different rule; see the article on base URL with and without /v1↗.

Step 4. First request

curl

curl https://api.zukko.pro/v1/chat/completions \
  -H "Authorization: Bearer fetch_your_key" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "grok-4.7",
    "messages": [
      {"role": "user", "content": "Explain in two sentences what an API is."}
    ]
  }'

The response has the text in choices[0].message.content and the number of tokens used in usage.

Python

from openai import OpenAI

client = OpenAI(
    base_url="https://api.zukko.pro/v1",
    api_key="fetch_your_key",
)

resp = client.chat.completions.create(
    model="grok-4.7",
    messages=[{"role": "user", "content": "Explain in two sentences what an API is."}],
)
print(resp.choices[0].message.content)
print(resp.usage)

Install with pip install openai. It is better to take the key from an environment variable: OPENAI_API_KEY and OPENAI_BASE_URL=https://api.zukko.pro/v1, then plain OpenAI() is enough in code.

Streaming

stream = client.chat.completions.create(
    model="grok-4.7",
    messages=[{"role": "user", "content": "List three ways to speed up a website."}],
    stream=True,
)
for chunk in stream:
    if chunk.choices:
        print(chunk.choices[0].delta.content or "", end="", flush=True)

The if chunk.choices: check is needed because the last chunk carries only usage and an empty choices list.

Image input

All three Grok models in the catalogue accept image input. The format is the same as OpenAI's:

resp = client.chat.completions.create(
    model="grok-4.7",
    messages=[{
        "role": "user",
        "content": [
            {"type": "text", "text": "What is in the picture?"},
            {"type": "image_url", "image_url": {"url": "https://example.com/photo.jpg"}},
        ],
    }],
)

What Grok costs through the gateway

Assume 1,000 requests, each with 2,000 input and 500 output tokens. In total 2M input and 0.5M output. Formula: input (in millions) × input price + output (in millions) × output price.

• grok-4.7: 2 × 0.175 + 0.5 × 0.513 = 0.35 + 0.2565 = $0.6065.
• grok-4.6: 2 × 0.138 + 0.5 × 0.4 = 0.276 + 0.2 = $0.476.
• grok-4.5: 2 × 0.113 + 0.5 × 0.338 = 0.226 + 0.169 = $0.395.

For reference, the official prices in the catalogue are the same for all three: input $3, output $15 per 1M. At those prices the same 1,000 requests would cost 2 × 3 + 0.5 × 15 = $13.5. We do not compare model quality: use the strength index and test on your own tasks. More on the middle model of the line is in the Grok 4.6 article↗.

Common errors

The general connection steps are in the quick start↗.

FAQ

How do I use the Grok API?

Create a key, set the base URL https://api.zukko.pro/v1 and a model name such as grok-4.7, and send a request in the Chat Completions format. Any OpenAI-compatible client or SDK will do.

Do I need a foreign card?

No. Access from Russia goes through a gateway: payment by SBP (the payment page adds an 8.5% fee on top), MasterCard or crypto.

How does Grok differ from Claude and GPT in context?

Grok in the catalogue has a 256K-token window, Claude and GPT have 1M. For ordinary dialogues it is enough; for large documents estimate the size in advance: send a trial request with part of the document and look at usage.prompt_tokens.

Can I call Grok in the Anthropic format?

Yes: any catalogue model can be called both in the OpenAI format and in the Anthropic format (POST /v1/messages).

To get started, create a key in the Zukko dashboard↗ and repeat the request from step 4.