Your Janitor AI proxy. About 8,000 DeepSeek V4.1 Flash messages for $10.

Paste a proxy URL, a model name and your key into Janitor AI and chat on DeepSeek, Gemini or Claude. Pay for what you use, no subscription needed, with credit that never expires.

Your proxy settings

Proxy URL: https://api.apisrouter.com/v1/chat/completions
Model: deepseek-v4.1-flash
API key: sk-your-apisrouter-key

What $10 buys

One message in a running chat: about 8,000 tokens of character card and history in, 400 out, priced without caching.

  • DeepSeek V4.1 Flash (deepseek-v4.1-flash): 8,000 messages for $10, $1.24 per 1,000 messages. The lowest cost here, so long chats stay cheap.
  • DeepSeek V4 Pro (deepseek-v4-pro): 2,000 messages for $10, $5 per 1,000 messages. DeepSeek's larger model, for replies that need more thought.
  • Gemini 3.8 Flash (gemini-3.8-flash): 1,600 messages for $10, $6 per 1,000 messages. Google's Gemini, for a different writing voice.
  • Claude Opus 5.5 (claude-opus-5-5): 300 messages for $10, $30.40 per 1,000 messages. Anthropic's Opus, for scenes where every line matters.

Set it up in one minute

  • Get a key: Top up from $10 below. The key and a login link arrive by email.
  • Open API Settings: In any chat on Janitor AI, open API Settings and switch from the default model to Proxy.
  • Fill in the three fields: Paste the proxy URL, the model ID and your key exactly as shown. The URL must end in /chat/completions.
  • Save and say hello: Save, refresh the chat and send a short message.

If it does not answer

  • model_not_found: Use the exact model ID, such as deepseek-v4.1-flash, not the model's display name.
  • 404: If the error does not say model_not_found, the proxy URL stops at /v1: add /chat/completions to the end.
  • 401: The key is incomplete or has a space at either end. Paste it again.

Two ways to pay

Both fill one balance, spent as you chat. Pay as you go from $10 with no account, with +30% on the first top-up.

FAQ

How much does a Janitor AI proxy cost?

Through APIsRouter, $10 buys about 8,000 messages on DeepSeek V4.1 Flash and about 300 on Claude Opus 5.5, for messages the size shown above; longer chats cost more per message. Top up from $10, with +30% on the first top-up.

Can I try it for free?

Yes. Sign in with your email or Google and claim a $2 trial credit from your dashboard, no card needed. When it runs out, top up from $10.

Which model is best for Janitor AI?

Start with DeepSeek V4.1 Flash: about 8,000 messages for $10. Switch to DeepSeek V4 Pro or Claude Opus 5.5 for scenes that need more care, and back for everyday chat. The model is one field in Janitor AI's settings.

How do I use DeepSeek on Janitor AI?

Set the model to deepseek-v4.1-flash or deepseek-v4-pro with the proxy URL above. DeepSeek's own API works too, but it charges double on weekday peak hours; APIsRouter charges one price at every hour.

Why do I get a 404 or 401 error?

A 404 that says model_not_found means the model ID is not exact; any other 404 means the proxy URL stops at /v1 instead of ending in /chat/completions. A 401 means the key is incomplete or has a stray space.

Does it work in SillyTavern and other chat apps?

Yes. Chat apps that take an OpenAI-compatible URL and key work. Most want the base URL, https://api.apisrouter.com/v1, rather than the full chat completions URL Janitor AI asks for.

Do unused credits expire?

No. Credit stays in your balance until you use it, and requests stop when it runs out, so there are no surprise bills.