[Napster3301]

TARTARUS AI
Three models on our own GPUs. Reduced refusal training. Crypto only.

---

Think ChatGPT, without the refusals. Same kind of assistant - chat, code, research, writing - except a direct question gets a direct answer instead of a lecture, a hedge, or a watered-down version of what you asked for. We run our own models on our own hardware, which is why we can do that.


Three models, picked per message

• Fast - quick answers for everyday questions. No reasoning step.
  


• Balanced - thinks a little before answering. A good default.
  


• Deep - thinks before answering. Best for hard problems.
  

Balanced and Deep show their reasoning separately, so you can read the work instead of trusting the conclusion.

What comes with it

• Reduced refusal training - our models answer from what they know, not from a list of subjects they learned to decline.
  


• Cited web search - the source sits on the claim it supports.
  


• Context - 200,704 tokens on Pro and Ultra, 131,072 on Standard, 32,768 on Basic.
  


• OpenAI-compatible API and CLI, Pro and up - one base URL, 10 keys.
  

Tiers - per month, prepaid

• $20 Basic - web only, Fast, 32K context
  


• $40 Standard - Fast + Balanced, 131K context, no API
  


• $60 Pro - all three models, 200K context, API + CLI
  


• $100 Ultra - Pro at twice the volume
  

10% off three months prepaid, 20% off twelve. USDT (TRC20/ERC20), Bitcoin, Ethereum, Litecoin, Tron, Monero. No cards. Verified email required.

---

tartarusai.chat - what's off-limits

We read this forum too - ask us anything, including which layer of the stack actually decides whether a model declines you.
Start Contract

[UnmarkedPad]

Hey, I'm looking to set up an AI agent on my VPS, mainly to be an assistant and help record findings, but also one that can code and help develop. I can read code pretty well, but my memory is so bad I forget things instantly, so looking for something to be my dictionary if you like.

I'm still fairly new to doing this and as it's in my VPS and all CLI based, would I be able to just subscribe to like a certain amount of tokens or time to use the API?

[Napster3301]

Hey, I'm looking to set up an AI agent on my VPS, mainly to be an assistant and help record findings, but also one that can code and help develop. I can read code pretty well, but my memory is so bad I forget things instantly, so looking for something to be my dictionary if you like.

I'm still fairly new to doing this and as it's in my VPS and all CLI based, would I be able to just subscribe to like a certain amount of tokens or time to use the API?

It should work for those purposes. Please just keep in mind that API support is available starting with the Pro plan.