Unrestricted AI chat through a developer-controlled API
Unrestricted AI chat served through an OpenAI-compatible API: consistent behavior across sessions, no per-request refusals, and project-scoped keys.
Unrestricted AI chat is about consistency: the same class of conversation behaves the same way across sessions, without per-request refusals on authorized workloads.
Provider-hosted chat products apply silent, shifting filters; an API-first approach puts conversation behavior under your control.
abliteration.ai serves unrestricted chat behavior through standard chat completions with streaming, scoped by project keys and quotas.
Unrestricted AI chat through a developer-controlled API
Unrestricted AI chat is chat served by a model endpoint with reduced provider-side refusal behavior, where the application owns session policy, keys, and quotas.
- Per-request refusals on valid workloads break multi-turn flows and analyst tooling.
- Session consistency matters for red-team scripts, eval harnesses, and support automation.
- Project-scoped keys let each chat workload carry its own quota and audit trail.
- 01Call /v1/chat/completions with model: abliterated-model; enable stream: true for interactive UX.
- 02Issue one key per chat product or environment to isolate usage.
- 03Apply your conversation policy in-app, or route end-user traffic through Policy Gateway.
import OpenAI from "openai";
const client = new OpenAI({
apiKey: process.env.ABLIT_KEY,
baseURL: "https://api.abliteration.ai/v1",
});
const stream = await client.chat.completions.create({
model: "abliterated-model",
stream: true,
messages: [
{
role: "system",
content: "You are an analyst assistant inside a governed workspace.",
},
{
role: "user",
content: "Walk through this incident timeline and flag anomalies.",
},
],
});
for await (const chunk of stream) {
process.stdout.write(chunk.choices[0]?.delta?.content ?? "");
}Serve unrestricted chat under your own policy
Create a project key and stream your first unrestricted chat completion.
Get an API keyChat workloads teams run unrestricted
- Red-team conversation scripts that must run to completion.
- Analyst copilots working over sensitive incident data.
- Moderation tooling that reviews borderline content to classify it.
- Regression suites that replay recorded sessions against new model versions.
Controls that ship with the API
- Project-scoped API keys for each chat surface.
- Per-project quotas with auto-reload for long-running workloads.
- Zero prompt retention by default.
- Optional Policy Gateway with explicit decision codes and audit logs.
Frequently asked questions.
Does unrestricted chat keep history across sessions?
The API is stateless like any chat completions endpoint. Your application manages conversation history and decides what to send with each request.
Will the model refuse mid-conversation?
Provider-side refusals are reduced, so authorized workloads behave consistently. Your application policy still governs what is served to end users.
Can I stream responses?
Yes. Set stream: true and consume server-sent events with any OpenAI-compatible client.
How do I keep unrestricted chat compliant?
Use project-scoped keys, per-project quotas, audit logs, and Policy Gateway rules for end-user-facing traffic.
Which endpoints support chat workloads?
The OpenAI-compatible /v1/chat/completions, /v1/responses, and /v1/messages endpoints, all with model: abliterated-model.