Zed AI in Russia: Claude, DeepSeek and GPT in the Zed editor, no VPN or foreign card
Zed is a fast, open-source code editor for macOS, Windows and Linux with a built-in AI agent. Zed’s own models come with the Pro plan at $10 a month, paid by card. But AI also works on the free Personal plan — with your own API key. Create a provider in Zed for https://www.ruapi.ai/v1, enter a RuAPI key, and the Agent Panel, Inline Assistant and commit message generation run on Claude, DeepSeek, GPT and Gemini. No VPN, no Zed sign-in, pay in USDT.
Updated:
from openai import OpenAI
client = OpenAI(
api_key="sk-...",
base_url="https://www.ruapi.ai/v1",
)
resp = client.chat.completions.create(
model="deepseek-flash",
messages=[{"role": "user", "content": "Hello"}],
)
print(resp.choices[0].message.content)You can check the key with the cURL tab right in Zed’s built-in terminal. Later, when adding the provider, the address https://www.ruapi.ai/v1 goes into API URL and the key into API Key.
Why AI in Zed does not just work from Russia
Without Pro, the Zed agent only runs on your own key.
The models Zed sells and its edit predictions come with the Pro plan at $10 a month, and paid plans are billed by card through Stripe (zed.dev/pricing, zed.dev/docs/account/billing, 3 Nov 2026). The free Personal plan gives you an unlimited agent, but with your own API keys. A direct Anthropic or OpenAI key is out of reach from Russia, since neither company serves it (anthropic.com/supported-countries, developers.openai.com/api/docs/supported-countries, 3 Nov 2026). A RuAPI key, on the other hand, is ready as soon as you register.
Every vendor has its own provider and its own bill in Zed.
Zed’s settings list Anthropic, OpenAI, Google AI, DeepSeek and other providers separately, and each needs an account with its own company. With RuAPI, one OpenAI-compatible provider and one key are enough: Claude, GPT, Gemini, DeepSeek, Qwen, GLM and Kimi are all in the Agent Panel model picker. Each request’s price is then visible in Usage Logs in the RuAPI Console.
The agent spends tokens at every step.
Zed Agent reads files, searches the project and runs commands on its own, and at every step the model gets the whole task context again. So give the Zed key a budget. Housekeeping work — thread titles and commit messages — can go to a separate cheap model via thread_summary_model and commit_message_model in settings.json, such as deepseek-flash at $0.168 per 1M input tokens.
How to connect RuAPI to Zed: 3 steps
Written for Zed 1.26 stable on macOS, Windows and Linux. You do not need to sign in to Zed: per the docs, AI works without signing in when you bring your own API keys (zed.dev/docs/authentication, 3 Nov 2026).
Sign up for RuAPI and top up
An account at www.ruapi.ai takes an email, Google or Yandex login and no phone number. USDT is the most convenient way to pay: from 5 USDT over any of TRC20, BEP20, TON or Polygon (USDC is welcome too). Where and how to buy USDT for rubles through SBP is explained at docs.ruapi.ai/en/buy-usdt. A card from a foreign bank is the alternative.
Trying the Zed agent costs nothing at first: new accounts usually come with $0.05 for claude-haiku-5-5, deepseek-flash or gpt-6-luna; access to the other models comes with your first top-up.
Create a separate key for Zed
Console → API Keys → “Create API key”. In the Credit limit block, turn off “Unlimited” and enter an amount: even if the agent gets stuck in a loop, it cannot spend more. Zed keeps the key in the system keychain, not in settings.json, so you can safely commit your settings file to a dotfiles repo.
Add RuAPI as a provider in Zed
- Open Settings → AI → LLM Providers with the agent: open settings command from the command palette (Cmd+Shift+P on macOS, Ctrl+Shift+P on Windows and Linux). Click Add Provider and choose OpenAI under Compatible APIs.
- Provider Name — ruapi: Zed builds the environment variable RUAPI_API_KEY from this name, and you can pass the key that way too. API URL — https://www.ruapi.ai/v1; without the /v1 Zed would call the wrong address. API Key — your sk-… key.
- In the Models block click Add Model. Model Name takes the model ID exactly as written on the pricing page, e.g. deepseek-flash. Max Tokens — the context window, 1000000 for deepseek-flash. Max Output Tokens can stay at 32000. Keep Supports tools checked: without tools the agent cannot read or edit files. For DeepSeek, also check Uses max_tokens for output limit, and after saving turn on "interleaved_reasoning" in settings.json — see the FAQ below.
- Click Save Provider. Open the Agent Panel (the ✨ icon in the status bar), pick deepseek-flash under ruapi in the model list and ask something about your project. If a new line appears in Usage Logs in the RuAPI Console, the setup works.
RuAPI addresses and keys: docs.ruapi.ai/en/quickstart · tool calling: docs.ruapi.ai/en/function-calling
What to set in Zed
OpenAI provider — all models
API URL: https://www.ruapi.ai/v1
API Key: sk-…
Model Name: deepseek-flash
Max Tokens: 1000000
env variable: RUAPI_API_KEY
Anthropic provider — Claude only
API URL: https://www.ruapi.ai
Model Name: claude-sonnet-5-5
"prompt_caching": true
"max_output_tokens": 32000
Payment
USDT / USDC from 5 USDT
foreign-issued cards
no subscription
Model prices for the Zed agent
Models that perform well in the Agent Panel and Inline Assistant, priced in USD per 1M tokens according to the pricing page. No subscription: only the tokens the editor spends are charged.
| model | input $/M | output $/M |
|---|---|---|
| deepseek-flash | $0.17 cache read $0.0120 | $1.44 |
| glm-5.3 | $1.44 cache read $0.36 | $5.02 |
| gpt-5.4-mini | $0.90 cache read $0.0900 | $5.40 |
| claude-haiku-5-5 | $0.12 cache read $0.0120 >100K: $0.60 | $0.60 >100K: $3.00 |
| gemini-3.8-flash | $0.90 cache read $0.0900 | $4.50 |
| gpt-5.4 | $3.00 cache read $0.30 >272K: $6.00 | $18.00 >272K: $27.00 |
| claude-sonnet-5-5 | $2.40 cache read $0.12 | $12.00 |
Input covers new tokens going in, “cache read” the part of the input that came from cache, output the generated reply. At every step the agent resends files it has already read and the thread history, so caching saves a lot here. GPT and DeepSeek pick up a repeated request prefix from cache automatically. Zed’s OpenAI provider adds no cache_control marks, and the Anthropic provider adds them only with "prompt_caching": true; otherwise RuAPI adds them itself when the next agent step comes within 5 minutes (docs.ruapi.ai/en/prompt-caching).
The “>100K” row, shown in grey, is the vendor’s rate for prompts above the threshold; a request that exceeds it is paid entirely at that rate.
Free Zed with RuAPI or a Zed Pro subscription
| Zed Personal + RuAPI | Zed Pro | |
|---|---|---|
| Plan price | $0, unlimited editor and agent | $10 a month, $5 of tokens included |
| Paying for models | Only the tokens you use; USDT from 5 USDT or a foreign-issued card | Beyond $5 — vendor list price +10%; card via Stripe |
| Zed account | Not needed | GitHub sign-in required |
| Edit predictions (Zeta) | No; Copilot, Codestral or a local model instead | Unlimited |
Based on zed.dev/pricing and zed.dev/docs (Plans & Pricing, Billing, Authenticate with Zed), checked 3 Nov 2026. Pro also has a 14-day trial with $5 of GPT-6 Luna usage.