Neural Network APIs in Russia: Protocols, Keys, and Your First Request

APIs, web subscriptions, and someone else's account are different things

API access means an application sends an HTTP request with your key to a provider endpoint. A web subscription provides access to a specific service's interface and does not necessarily include the same API balance. Buying or sharing someone else's account adds security risks and is usually unnecessary for API integration.

With BetterToken, each user has their own account and API key. BetterToken is not ChatGPT Plus, Claude.ai, or an official OpenAI or Anthropic subscription, and it does not sell shared accounts.

How to choose a protocol

Open the documentation for the tool or SDK and find the exact interface name.

Choose OpenAI-compatible if the client expects the OpenAI SDK, Chat Completions, Responses API, or a field such as OPENAI_BASE_URL. Choose Anthropic-compatible if the client constructs Messages API requests and uses ANTHROPIC_BASE_URL or x-api-key.

The model name does not determine the protocol. The client must be able to construct the same contract that the endpoint accepts.

BetterToken uses different base addresses:

OpenAI-compatible Base URL:  https://www.bettertoken.ai/v1
Anthropic-compatible Base URL: https://bettertoken.ai

The first address already includes /v1. The second is the Base URL for Anthropic-compatible clients; the full HTTP Messages path adds /v1/messages.

Base URL versus full request path

The Base URL defines the API root for an SDK or ready-made tool. You need the full path for a direct HTTP request.

An OpenAI-compatible full path may look like this:

https://www.bettertoken.ai/v1/chat/completions

Anthropic Messages uses:

https://www.bettertoken.ai/v1/messages

If an application field is named Base URL, do not paste the full path unless its documentation explicitly tells you to. Otherwise, the client may append the resource path a second time and return 404.

How to store an API key

Use an environment variable for a local test. A provider-neutral variable name helps prevent configuration mix-ups:

export BETTERTOKEN_API_KEY="your_api_key_here"
export BETTERTOKEN_MODEL_ID="your_current_model_id"

Do not put a real key in source code, .env.example, an issue, a prompt, a screenshot, or a shell command that will be saved in shared history. In production, use the platform's secret manager.

Minimal OpenAI-compatible request

Check the current endpoint and format in the API reference first. Here is a safe template with placeholders:

curl https://www.bettertoken.ai/v1/chat/completions \
  -H "Authorization: Bearer $BETTERTOKEN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "'"$BETTERTOKEN_MODEL_ID"'",
    "messages": [{"role": "user", "content": "Reply with API_OK"}],
    "max_tokens": 16
  }'

Do not include -v in a public log: verbose mode can expose sensitive headers.

Minimal Anthropic-compatible request

Get the protocol version and supported headers from the current API reference. The Messages structure is different:

curl https://www.bettertoken.ai/v1/messages \
  -H "x-api-key: $BETTERTOKEN_API_KEY" \
  -H "anthropic-version: CURRENT_SUPPORTED_VERSION" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "'"$BETTERTOKEN_MODEL_ID"'",
    "max_tokens": 16,
    "messages": [{"role": "user", "content": "Reply with API_OK"}]
  }'

CURRENT_SUPPORTED_VERSION is a placeholder. Do not copy a historical value without checking the current documentation.

If your task specifically involves a Claude-compatible connection, check the key, billing, and protocol boundaries on the Claude API page before configuring the client, then return to the minimal request above.

What to check after the response

The minimal test passed if several signals line up:

• The HTTP status indicates a successful request;
• the response names the expected Model ID or its documented mapping;
• usage is present if the selected contract returns it;
• the BetterToken Dashboard shows a record with the model, status, input/output/cache tokens, and the corresponding charge.

Do not hard-code the model price in an article: the catalog and prices change. Use the current model page.

Troubleshooting errors

401

Check the key, stray spaces, and authentication method. Do not swap a Bearer token for x-api-key, or vice versa, unless the protocol requires it.

404

Compare the Base URL with the full path. A duplicated /v1 or /chat/completions is a common cause.

model not found

Copy the current Model ID and make sure it is available for the selected key and protocol. A model's marketing name may differ from its API ID.

Timeout or TLS error

Separate local networking, proxy, and certificate issues from the API response first. A short curl request without an SDK helps establish whether the request reaches the endpoint. Do not disable TLS verification as a permanent workaround.

A practical sequence

Identify the client contract, store the key as a secret, use the correct Base URL, send one short request, and check the Dashboard. Only then enable streaming, tools, long context, or an agent workflow: each layer introduces its own event formats and failure modes.

Next step: OpenAI-compatible APIs and OpenRouter alternatives

To configure your own key and an OpenAI-compatible route, open the OpenAI API page. This is the BetterToken API with a compatible interface, not an official OpenAI key. If your current integration uses OpenRouter, compare the requirements on the OpenRouter alternatives page before changing a production route.