Skip to main content
Continue is no longer developed: the Continue team joined Cursor. The continuedev/continue repository is read-only and version 2.0.0 is the final release: no new versions, fixes or support for new models. If you already have Continue installed, you can keep connecting it to RuAPI with the steps below. For new users we recommend Kilo Code (VS Code and JetBrains) or Cline (VS Code).
Continue is an open-source AI coding assistant for VS Code and JetBrains: chat about your codebase, edit selected code, or let agent mode make changes to files on its own. Models are defined in a config.yaml file. RuAPI gives you access to Claude, GPT, Gemini, DeepSeek, GLM, Qwen and Kimi with a single API key, and works from Russia without a VPN. You top up your balance in USDT; no foreign-issued card required.

Before you start

1

RuAPI API key

Sign up at www.ruapi.ai and top up your balance (how to top up). Then go to Console → API Keys and click Create API key. Keys start with sk-.
2

The Continue extension

Install Continue from the VS Code Marketplace or the JetBrains Marketplace (search for “Continue”). The JetBrains Marketplace still has the old 1.0.67 version, and the developers’ own README recommends the Continue CLI instead of the JetBrains plugin. If you work in JetBrains, go straight to Kilo Code.
No bank card needed: you can top up your balance in USDT. If you don’t have USDT yet, follow our step-by-step guide to buying it with rubles. Foreign-issued cards are accepted too.

Where to find the config file

Open the Continue panel, click the configuration selector at the top of the panel and choose the local configuration (Local Config) — this opens config.yaml. The file is located at:
  • macOS and Linux: ~/.continue/config.yaml
  • Windows: %USERPROFILE%\.continue\config.yaml

Ready-to-use configuration

Replace the contents of config.yaml with the example below, then put your key in place of sk-YOUR_KEY (in both places):
Key points:
  • Claude uses provider: anthropic. Requests then use Claude’s native format, and promptCaching: true enables prompt caching: context that gets re-sent costs at least 10× less. Claude also works with provider: openai, but Continue then sets no cache markers, so only RuAPI auto-caching applies (on the first turn of a new conversation at most the system prompt and tools are cached, and after a gap of more than 5 minutes the cache expires).
  • All other models use provider: openai. This is how you connect DeepSeek, GPT, Gemini, GLM, Qwen and Kimi — just change model and name.
  • apiBase is https://www.ruapi.ai/v1 for both providers, including the trailing /v1.
  • capabilities: tool_use makes the model available in agent mode. Continue detects this by itself for Claude with provider: anthropic and for GPT. For Claude with provider: openai, and for DeepSeek, GLM, Qwen, Kimi, Gemini and other models, set capabilities: tool_use explicitly.
You’ll find the exact model names on the pricing page. Enter them exactly as listed, including any prefix and capitalization (e.g. kimi/kimi-k3).

Adding more models

To add a model, copy an entire block and change name and model. For example:
Continue only switches GPT-5 to the newer Responses API when connected directly to api.openai.com. With the RuAPI address, GPT requests go through /v1/chat/completions, and there’s nothing extra to set.
For the newest Claude models (Opus 4.7 and later, plus all fifth-generation models), don’t add reasoning or reasoningBudgetTokens to defaultCompletionOptions: these models don’t support the legacy way of enabling reasoning and will return a 400 error.

Verify the setup

Open the Continue chat (Cmd + L on macOS, Ctrl + L on Windows and Linux), pick a model at the top — for example, “Claude Sonnet 5.5 (RuAPI)” — and ask a question about your code. Then open the RuAPI Console → Usage Logs: you’ll see an entry with the model and the cost of the request.

Troubleshooting

  • apiKey must contain your full RuAPI key (sk-...).
  • apiBase must be https://www.ruapi.ai/v1, including the trailing /v1.
The model value must match the name on RuAPI exactly — copy it from the pricing page. Claude models work with either provider: anthropic or provider: openai; all other models work only with provider: openai.
Add capabilities with the value tool_use to the model block, as in the DeepSeek example.
YAML is indentation-sensitive: use spaces, not tabs, and keep the structure of the example. Continue reloads the file automatically after you save it.
Code autocomplete (the autocomplete role) in Continue is designed for dedicated autocomplete models. Regular chat models are slow and expensive in that role, so the example leaves it out — RuAPI is best suited for chat, edits and agent mode.

What’s next