Open LLMs,
abliterated.
Models that will actually help with offensive and defensive security work — for red teams, SOC analysts and researchers. One API, billed per token, and your prompts are never written to disk.
Open-weight models,
priced lower
One endpoint that speaks both the OpenAI and the Anthropic formats. Top up once, pay for the tokens you use, and nothing you send is kept.
Keeps your current client
Change the base URL to , pick a model id, done. Code written for the OpenAI or Anthropic SDKs — and the agents built on them — keeps working as is.
Edited weights, no prompt tricks
The refusal behaviour is removed from the model’s weights themselves. We don’t inject a hidden system prompt, so long sessions don’t drift back into refusing. How abliteration works
Your prompts aren’t kept
Prompt and completion text is passed through and dropped — never logged, never used for training. We only keep the counters billing needs. Privacy policy
The current line-up
Live prices from our catalogue. Click an id to copy it, or open a row for the full spec sheet.
Three steps to
your first call
- 01Create an account and top up
Prepaid balance in USD, funded with crypto. No subscription.
- 02Generate an API key
Keys start with lj_. Give each one its own budget or expiry date.
- 03Point your SDK at us
Same request format you already use. Streaming works out of the box.
Questions
Still unsure? Ask us on Telegram — a person answers.
Your first request is
a minute away
Sign up, add credit, paste a key. That’s the whole onboarding.