abliteratly is an OpenAI-compatible API for 8 uncensored models and 178 frontier models. It is built for people who want one key, one balance and a straight answer about what the service does.
Access to language models through a chat-completions endpoint. The uncensored tier is the reason the service exists; the frontier tier exists so you do not need a second account for the parts of a workload that need a mainstream model. The catalog is deliberately narrow: chat completions, streaming, tools, JSON mode and vision where supported. No image generation, embeddings or audio.
Frontier models are passed through at the upstream list price. The margin is on the uncensored tier. There is no subscription, no monthly minimum and no renewal. You buy credit with crypto and spend it per token. Credits do not expire.
We do not store prompt or response bodies. We store account, key, payment and request metadata needed to run the service. Upstream model providers have their own retention policies, which vary by model. If a provider's policy matters for your workload, check the model before you use it.
No model is guaranteed to answer every prompt. Uncensored means fewer built-in refusals, not better judgment or accuracy. Model availability can change. The service is provided as available, not as an SLA. Read the terms, privacy policy and acceptable use policy.
Start with a small test set from your own workload. Compare output quality, tool reliability and cost against the model you use today. The pricing page has a calculator; the model directory has live rates. If the numbers do not work for your workload, the service is not the right fit.
A small team building a focused API for uncensored and frontier models. The service is operated independently.
No. The status page shows observed availability from real requests, but it is not an SLA or a contractual guarantee.
Use the contact channel on the status page. There is no email account requirement, but support still needs your account ID and request details.