Windsurf in Russia: a Windsurf AI alternative with no VPN
Since June 2, 2026, Windsurf IDE has been called Devin Desktop and belongs to Cognition, the company behind the Devin agent. The Cascade agent no longer runs, models draw on a daily and weekly subscription quota, and the built-in agent has no field for your own API address (docs.devin.ai, 7 Nov 2026). What you can do is install Cline or Kilo Code in the editor itself — or move to VS Code, the terminal or Zed — and connect them to RuAPI: Claude, GPT, DeepSeek and Grok with one key, no VPN, paying in USDT only for the tokens you use.
Updated:
API Provider: Anthropic
Anthropic API Key: sk-...
Use custom base URL: https://www.ruapi.ai
Model: claude-sonnet-5-5
These are the fields in Cline’s settings. The extension installs both in Devin Desktop (formerly Windsurf) from Open VSX and in regular VS Code. Connect Claude through the Anthropic provider and other models through OpenAI Compatible; replace sk-… with your RuAPI key.
What happened to Windsurf, and why it’s awkward from Russia
Windsurf is now Devin Desktop, and Cascade is gone.
Cognition bought Windsurf in July 2025, and on June 2, 2026 the editor was renamed Devin Desktop through a regular update (docs.devin.ai/desktop/devin-desktop-faq). New Cascade conversations no longer start and old ones no longer open: the local agent is now Devin Local (docs.devin.ai/desktop/cascade, 7 Nov 2026). Since you have to get used to a new agent anyway, it makes sense to pick one that works with any model and isn’t tied to a subscription.
The subscription is hard to pay for from Russia.
Free is $0 with a small quota, Pro is $20 a month, Max is $200 (devin.ai/pricing, 7 Nov 2026). Since March 2026 credits have been replaced by a daily and weekly quota: on Free you wait for the reset, on paid plans you buy extra usage at API prices (docs.devin.ai/desktop/accounts/quota). Payment is by card, Apple Pay, Google Pay or Link through Stripe (docs.devin.ai/desktop/accounts/usage), and Russian bank cards don’t work abroad. Paid plans have no free trial.
You can’t plug in your own key.
BYOK arrived in Windsurf 1.9.2 (22 May 2025) and accepted only an Anthropic key — for Claude Sonnet 4 and Opus 4 in Cascade (docs.devin.ai/desktop/changelog). The BYOK notes say nothing about a custom API address like OpenAI-compatible providers offer, and the Devin Desktop models page doesn’t mention your own key at all now (docs.devin.ai/desktop/models, 7 Nov 2026). So a RuAPI key can’t go into the built-in agent — you need an agent that works with any API address.
Moving from Windsurf to an agent on RuAPI: 4 steps
The example uses Cline, which installs both in Devin Desktop and in VS Code. Kilo Code, OpenCode, Zed and other tools are set up in a similar way — detailed pages are listed under “See also”.
Sign up for RuAPI and top up
Sign up at www.ruapi.ai with email, Google, Yandex or Telegram. Top up in USDT from 5 USDT on TRC20, BEP20, TON or Polygon; USDC is accepted too. To buy USDT with rubles via SBP, see docs.ruapi.ai/en/buy-usdt. Foreign-issued cards are accepted too.
A new account usually gets $0.05 right away — enough to try the agent on claude-haiku-5-5, deepseek-flash or gpt-6-luna. The other models unlock after your first top-up.
Create a key with its own budget
Console → “API keys” → “Create key”. Instead of a quota that resets daily and weekly, you set the limit: turn off “Unlimited” in the “Budget” block and enter an amount — the key will never spend more than that.
Install Cline where it suits you
- Stay in Devin Desktop: the Extensions panel pulls packages from an Open VSX mirror by default (docs.devin.ai/desktop/advanced). Search for Cline — on open-vsx.org it’s saoudrizwan.claude-dev, version 4.1.26 from 26 Oct 2026. Themes, keybindings and your other extensions stay as they are.
- Move to VS Code: Devin Desktop is built on VS Code — version 3.12.17 from 6 Nov 2026 is based on VS Code 1.141.2 (docs.devin.ai/desktop/changelog) — so the interface is almost the same. Cline is in the VS Code Marketplace.
- The Windsurf JetBrains plugin has moved to maintenance mode: no new features are coming (docs.devin.ai/desktop/devin-desktop-faq). In IntelliJ IDEA, PyCharm and other JetBrains IDEs, Kilo Code takes its place. For the terminal, there’s OpenCode, Claude Code or Codex.
Connect the models and move your project rules
- Claude: API Provider — Anthropic, key sk-…, turn on Use custom base URL and enter https://www.ruapi.ai (no /v1), model — claude-sonnet-5-5. Prompt caching works in this format.
- DeepSeek, GPT, Grok and the rest: API Provider — OpenAI Compatible, Base URL https://www.ruapi.ai/v1, the same key, Model ID — the exact name from the pricing page, e.g. deepseek-flash.
- Rules: Cline detects .windsurfrules on its own and lists it in the Rules panel (docs.cline.bot/customization/cline-rules); copy rules from the .windsurf/rules folder into .clinerules/. Turn workflows from .windsurf/workflows into Cline skills: .cline/skills/<name>/SKILL.md, invoked with /<name> in chat (docs.cline.bot/customization/skills).
- Ask the agent a question about your project — a new line with the cost of the answer will appear in the RuAPI Console request log.
Full Cline setup: docs.ruapi.ai/en/integrations/cline · addresses and keys: docs.ruapi.ai/en/quickstart · caching: docs.ruapi.ai/en/prompt-caching
What you used in Windsurf — and what replaces it
Cascade and Devin Local → an open agent
Cline: Plan and Act modes
Kilo Code: VS Code, JetBrains, terminal
OpenCode, Claude Code, Codex in the terminal
Rules and workflows → project files
.windsurfrules — Cline reads it as is
.windsurf/rules → .clinerules/
workflows → .cline/skills/
AGENTS.md for OpenCode and Codex
Subscription quota → RuAPI balance
no daily or weekly limits
pay only for tokens
a budget for each key
USDT from 5 USDT, balance never expires
Model prices instead of the Windsurf quota
The models people most often connect to coding agents in the editor. Prices are per 1M tokens in US dollars, from the RuAPI pricing page. There are no quotas: you’re charged exactly for the tokens the agent used.
| model | input $/M | output $/M |
|---|---|---|
| deepseek-flash | $0.17 cache read $0.0120 | $1.44 |
| gpt-6-luna | $0.12 cache read $0.0120 >272K: $0.24 | $0.60 >272K: $0.90 |
| claude-haiku-5-5 | $0.12 cache read $0.0120 >100K: $0.60 | $0.60 >100K: $3.00 |
| grok-4.7 | $2.40 cache read $0.60 >200K: $4.80 | $7.20 >200K: $14.40 |
| claude-sonnet-5-5 | $2.40 cache read $0.12 | $12.00 |
| gpt-5.4 | $3.00 cache read $0.30 >272K: $6.00 | $18.00 >272K: $27.00 |
Compared with Windsurf Pro at $20 a month (devin.ai/pricing, 7 Nov 2026). Take one agent step: 30,000 input tokens, 24,000 of them from cache, and 1,000 output tokens. By our estimate it costs about $0.033 on claude-sonnet-5-5 (the 6,000 new tokens are written to the cache and billed at the cache-write price) and about $0.0027 on deepseek-flash, so $20 buys roughly 600 or 7,300 such steps. Reasoning tokens are billed as output, so reasoning models will cost more. Cognition doesn’t publish how many tokens the Pro quota includes, so an exact one-to-one comparison isn’t possible.
The greyed-out “>272K” row is the vendor’s rate for prompts above the threshold; if a request exceeds it, the whole request is billed at that rate.