Start now
Appearance
Site language
Price currency
Log inStart now
NewВывод реферального вознаграждения в USDT

Every model. One key.

An OpenAI-compatible gateway to GPT, Claude, Gemini, DeepSeek and dozens more models. One balance, per-request billing, pay in rubles without a VPN.

Start for freeDocumentation
  • No subscription
  • Pay in rubles
  • Request bodies never stored
  • and 61 more models in the catalog

Models from leading labs — behind one endpoint

  • OpenAI
  • Anthropic
  • Google
  • DeepSeek
  • xAI
  • Z-AI
  • Moonshot
  • Qwen
  • Xiaomi
  • MiniMax
  • +3
Migrate in a minute

Change one line. Keep the rest of your code.

Evrona speaks the OpenAI API. Official SDKs, LangChain, Cursor and anything with a custom base_url just work.

  • Same methods: chat, responses, images, streaming
  • Switch models with one parameter — no new keys or contracts
  • All spend from one USD balance
1 import os2 from openai import OpenAI3  4 client = OpenAI(−    api_key=os.environ["OPENAI_API_KEY"],5+    base_url="https://evrona.ru/v1",6+    api_key=os.environ["EVRONA_API_KEY"],7 )8  9 resp = client.chat.completions.create(10     model="openai/gpt-6-astra",11     messages=[{"role": "user", "content": "Hello!"}],12 )13 print(resp.choices[0].message.content)
Pick a model — only one line changes
GPT-6 Astra · OpenAIinput / output per 1M: $0.67 / $2.22
Inside

A gateway that accounts for every cent

One balance for every model

Top up once, spend on any provider. See where every request went.

balance$42.18$42.18 · example
  • OpenAIopenai/gpt-5.4 · 3 551 tok−$0.0020
  • Googlegoogle/gemini-3.8-flash · 1 178 tok−$0.0003
  • Anthropicanthropic/claude-fable-5.1 · 480 tok−$0.0021
  • Googlegoogle/gemini-3.7-flash · 3 656 tok−$0.0042
  • Anthropicanthropic/claude-fable-5-compressed-context · 2 447 tok−$0.0112

A receipt for every request

Tokens × price = charge. No rounding up to packages.

req_8f2a…c41openai/gpt-6-astra
1 240 × $0.67/1M$0.0008
380 × $2.22/1M$0.0008
charged$0.0016

Key limits and schedule

Spend limits per day, week, month. Keys work only in the hours you allow.

monthly limit$36.00 / $50.00
access openclosed

Private by default

We keep only billing metadata: model, tokens, time. Never prompt or response text.

Rubles, no VPN

Buy a top-up code in the store with rubles and redeem it in your account. Balance is kept in USD.

Store codeXXXX-XXXX-XXXX-XXXX$25

Provider down — request not lost

The gateway probes routes and moves traffic to a healthy one. You see it on the status page, not in error logs.

Model status →
Route A · primaryopenai/gpt-6-astraOutage
traffic moved automatically
Route B · backupopenai/gpt-6-astraServing traffic

* illustrative data

A catalog that grows without you lifting a finger

Open catalog
69models in the catalog right now
13providers behind one endpoint
3task types: text, images, video
1key and one balance for everything
Pricing

Pick a model — we’ll estimate the month

Catalog prices, no subscriptions. Pay only for what you send.

OpenAIGPT-6 Astra$0.67 / $2.22 per 1M
50
1 500
500
per month$3.19$0.0021 per request · $3.19
Start with this model

Ruble and euro prices use the exchange rate on the update date; billing is in USD. All models & pricing →

Who it’s for

One key — four ways to work

Write code, not integrations

  • Streaming, tool calling and JSON mode — as in the OpenAI API
  • Per-request spend in your account and via /v1/workspace/spend
  • A/B test models by changing one variable
Quick start in the docs →
~/console — zsh
$ export EVRONA_API_KEY=ev-••••a91f
$ python ask.py --model openai/gpt-6-astra "Summarize the text in 3 bullets"
→ POST https://evrona.ru/v1/chat/completions
200 OK · 1.38 s · 1 240 → 182 tok

• The gateway gives one key to every model
• One balance, per-request billing
• Pay in rubles without a VPN
Integrations

Works where you already work

Ready guides for code editors, agents and no-code platforms.

Claude CodeTerminal agentCodex CLITerminal agentAiderTerminal agentQwen CodeTerminal agentOpenCodeTerminal agentHermes AgentTerminal agentCursorAI editorClineIDE pluginn8nNo-code automationTelegram BotChat botOpenClawChat botLangChainLLM frameworkLlamaIndexRAG and data
All tools
Start

From sign-up to first response in three minutes

  1. 10:00

    Top up

    With a store code — it lands on your balance right away.

    Top-up code(opens in a new tab)
  2. 21:30

    Create a key

    Set a limit right away if you need one.

    ev-••••••••a91fCreate
  3. 33:00

    Send a request

    From code, terminal or your favorite tool.

    base_url = "https://evrona.ru/v1"

The short version

Will I need a VPN or a foreign card?
No. The service is reachable from Russia without a VPN, and a foreign card is not required: the balance is topped up with a code — buy it in the store, activate it in the dashboard, and the amount is credited immediately.
By what rule is the cost of a request charged?
Text models are billed by input and output tokens at the model price at the moment of the request. Media models carry a fixed price per generation. Every charge shows up in the transaction history and in the stats.
What does it take to connect through an OpenAI-compatible SDK?
Point the official SDK base_url at our service with the /v1 suffix and use your ev-… key. The rest of your code stays the same. Ready-made snippets for curl, Python and JavaScript are in the docs.
Are my prompts and the models' responses stored?
No. Neither the database, nor the logs, nor analytics keep request and response bodies. Only metadata remains in the stats: model, tokens, cost, time.
How do I put a ceiling on a key's spending?
On the key page, set a money limit for a day, a week or a month. As soon as the limit is reached, the gateway returns 429 and stops charging until the next window begins.

All questions →

First response in three minutes

Sign-up, key and top-up all happen online. No calls, no applications, no paperwork.

Start for free