model
claude-opus-4.6
v2026.05300+ modelssuccess rate 99.2%
one key
300+ models
one endpoint
OpenAI-compatible gateway to Anthropic, OpenAI, Google, xAI and dozens more. Change the base URL — and it just works. Pay in rubles via wu-gold.
# one key → 300+ models curl https://api.wuproj.com/v1/chat/completions \ -H "Authorization: Bearer wu-..." \ -H "Content-Type: application/json" \ -d '{ "model": "claude-opus-4.6", "messages": [{"role":"user","content":"hello"}] }'
✓ 200 OK·41ms·0.214 g·one endpoint, any provider
works out of the box with
Hermes
Janitor
Cursor
SillyTavern
OpenAI SDK
01 / 05request route
one request — any model
Your app sends a request to our /v1, we route it to the right provider and return the response. One format, unified billing in gold.
yourapp
OPENAI-COMPATIBLEWUPROJ/v1 · router · 99.97% uptime
Anthropicclaude-opus-4.6
route
you → wuproj → Anthropic
latency
41ms
cost
0.214 g
02 / 05why wuproj
/ 01
300+ models in one catalog
Claude, GPT, Gemini, Grok, DeepSeek, Kimi and dozens more — one endpoint instead of a zoo of keys.
/ 02
honest pay-per-token
Top up wu-gold and pay only for input and output tokens. No hidden per-provider subscriptions.
/ 03
drop-in for the OpenAI SDK
Set the base URL and key — LangChain, Cursor, SillyTavern and your backend keep working as before.
300+
models in the catalog
5providers
Anthropic · OpenAI · Google · xAI · DeepSeek …
99.2%
request success rate · last 24h
110₽ = 1 gold
pay in rubles, no commission
03 / 05ecosystem
more than a gateway
wu-gold works across all WuProj products — from the API to roleplay chats. One balance, one payment in rubles.
wu-gold · 1 = 110 ₽WuNest · roleplay chats
REC · CAM 02
00:01:22:41
04 / 05questions
Frequently asked questions
What is WuProj?
WuProj is an AI API and LLM proxy: one OpenAI-compatible key gives you access to 300+ language models from Anthropic, OpenAI, Google, xAI, DeepSeek and others. Instead of separate accounts, keys and invoices with each provider, you pay per token from a single wu-gold balance and call any model through one endpoint, https://api.wuproj.com/v1. It fits AI agents, coding, chatbots and roleplay.
Which models are available?
300+ models are available: Claude (Anthropic), GPT (OpenAI), Gemini (Google), Grok (xAI), DeepSeek and many more — all through one key and one endpoint. The model is set in the request's model field, e.g. claude-opus-4.6, gpt-5.4, gemini-3.1-pro or grok-4.20. The full catalog with the gold cost of each request is on the Models page.
How much does it cost and how do I pay?
You pay per token for exactly what you use from a wu-gold balance: 1 gold = 110 ₽, with no mandatory subscription. Top up in rubles. Each model's per-request price is shown in the catalog. For steady workloads there are plans.
Is WuProj compatible with the OpenAI API?
Yes — WuProj is a drop-in replacement. Set base_url to https://api.wuproj.com/v1, use a wu-… key and a model from the catalog; no code or SDK rewrite is needed. The request and response format is the same as OpenAI's (/v1/chat/completions, streaming, tools/function-calling, vision). It works with the official OpenAI SDK (Python, Node.js), as well as SillyTavern, Janitor, Cursor and any OpenAI-compatible client.
Can I use it for agents, coding and roleplay?
Yes, WuProj is built for all three. For AI agents and coding — access to top reasoning models (Claude, GPT, Gemini, Grok) through one key with function-calling and streaming. For roleplay (SillyTavern, Janitor) — the same models with pay-per-token and no subscription.
How is WuProj different from going to providers directly?
One key and one balance instead of separate accounts, keys and invoices per provider; payment in rubles; 300+ models at once; a single OpenAI-compatible format and automatic failover between providers. You're not locked to one vendor — switch models with a single line in the model field.
What is gold and why do I need it?
Gold is WuApi's internal billing currency (like the dollar for the US or the ruble for Russia — ours is Gold). Every model has a per-token price, and usage is deducted from your Gold balance as you go; a single request might cost ~0.05 Gold. 1 Gold = 110 ₽, top up in rubles, no mandatory subscription. Gold doesn't expire — it's just your balance, spent only when you make requests.
Which should a newcomer pick?
For the lowest price per request — Gemini 3.1 Pro / 3.5 Flash, billed in Gold. For maximum quality and the full lineup (Claude, GPT, Gemini, Grok, etc.) — any model in Gold. The best value is Gemini: cheapest and best-tuned, including caching for long sessions.
What does the WuStart subscription (300 ₽) give me?
WuStart is a WuApi plan: it gives 2 Gold on your balance plus a discount on buying more Gold. All platform models (including every Gemini) are available; usage is billed from the Gold balance.
Can I pay from outside Russia?
For now payments are Russia-only. Paying from other countries is technically doable — it needs arrangements with payment providers, which is a matter of time, and we plan to address it.
How do I connect JanitorAI (and why do I keep getting errors)?
In JanitorAI pick Proxy / Custom (OpenAI) and enter exactly: 1) Proxy URL — https://api.wuproj.com/v1/chat/completions (it must end with /chat/completions); 2) API key — your wu-… key from the dashboard, no spaces; 3) Model — any catalog ID, e.g. google/gemini-3.5-flash. A constant error is almost always a wrong URL (missing /chat/completions) or a typo in the key. SillyTavern and OpenAI SDKs use the base URL without the suffix: https://api.wuproj.com/v1. Also check your Gold balance (≈0.05 Gold per request).
05 / 05start
connect to wuproj
Get a key in a minute and send your first request. Pay in rubles, with no payment-system commission.