130+ AI models in one place
Chat in the browser and a key for Claude Code and Codex. No VPN, no foreign card. Pay in rubles only for answers: from 0.005 ₽.No VPN, no foreign card. Pay in rubles only for answers: from 0.005 ₽.
Chat in the browser and a key for Claude Code and Codex
Done
5 of 5 modules
You sent ≈ 46 tokens. The middleman charged 4,406 tokens.
We charged ≈ 46. Middlemen quietly add their own text to your request and charge you for it. We test the models with a control request four times a day, and every request's usage is in your dashboard.Their prompt. Your prompt_tokens.
typical reseller
request of 2026-10-23
Chargedprompt_tokens4,406tokens
your text ≈ 46your prompt ≈ 46
middleman's hidden textreseller's prompt
PRSLOY
Opus 5.5 · 2026-10-23
Chargedprompt_tokens≈ 46tokens
your text ≈ 46your prompt ≈ 46
no hidden extra textno injected prompt
| input | output | cache |
|---|---|---|
| 38 | 190 | 1.9 |
₽ per 1M
VERIFIEDA calculator without expiring subscriptions
≈0.116₽ per GPT-6.1 Sol answer to a large request
Example: 20,000 input tokens, 75% cached, 800 output tokens · 2026-10-27
20,000 × 0.1155 ₽
20,000 × 0.1155 ₽
500 × 0.1155 ₽
500 × 0.1155 ₽
Money doesn't expire at month end
One API for any task. 150 AI models for any task.models.
A four-drawer card catalogue organised by task. The drawer for the selected task is pulled out, with model cards rising from it.
- codeGPT‑6.1 Sol
- codeOpus 5.5
- textGemini 4
- textFable 5.1
- codeDeepSeek V4 Pro
- textGrok 4.7
- imagesGPT Image 2.5
- codeSonnet 5.5
- textQwen3.8 Max
- videoVeo 3.1
- codeKimi K3
- imagesNano Banana Pro
- textGLM‑5.3
- videoKling 3.0
- textHaiku 5.5
- textMiniMax M3
- codeDeepSeek 4.1 Flash
- videoSeedance 2.5
Connect
Chat right in the browser or by key in your own app
Make a Telegram bot that books haircut appointments, with buttons to pick a time
Done. Here is the bot's Python code and how to run it.
One base_url line and your agent runs through us
base_url = https://api.prsloy.online/v1
export ANTHROPIC_BASE_URL="https://api.prsloy.online"export ANTHROPIC_API_KEY="<key from the console>"export ANTHROPIC_DEFAULT_HAIKU_MODEL="claude-opus-5-5"export CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC=1claude --model claude-opus-5-5OpenAI API Key: <key from the console>Override OpenAI Base URL: https://api.prsloy.online/v1Add model: gpt-6.1-solfrom openai import OpenAI client = OpenAI( base_url="https://api.prsloy.online/v1", api_key="<key from the console>",)reply = client.chat.completions.create( model="gpt-6.1-sol", messages=[{"role": "user", "content": "Hello"}],)print(reply.choices[0].message.content)Ready-made builds
Not just AI models: ready, tested builds and settings for your tasks.
This is the next stage. For now, connect your own tool using the setup guide.
Open the setup guideEngineering drawing of a cassette deck: top-loading well, transport keys, and tape counter, with a tray of three skill cassettes in front. Inserting a cassette changes the machine diagram.
- Task
- Result
Questions
Do I need a VPN, a foreign card or my own Claude account?
No. Sign in with e-mail, pay in rubles. You do not need your own Claude or ChatGPT account, so there is nothing to ban.
Where do middlemen's tokens go?
They add their own hidden text to your request and give no cache discount. With us repeated chat history on GPT is 10× cheaper, and every request's usage is in your dashboard.
Is it hard to connect Claude Code or Codex?
No. The Setup page has ready settings for Claude Code, Codex and other tools: paste them and your agent works through us.
How much is it, and does money expire?
From 0.005 ₽ for a short answer on DeepSeek V4.1 Flash: 1,000 input and 800 output tokens. GPT-6.1 Sol example: 20,000 input tokens, 75% cached and 800 output tokens, about 0.116 ₽. Prices for all models are on the Models tab in the dashboard.