Obsidian X — OpenAI-compatible Inference API

One API.
Forty-seven models.
Zero rewrites.

Obsidian X is an inference gateway that speaks fluent OpenAI — streaming, tools, JSON mode and included. It's completely token-based: grab credits from our Discord, point your SDK at us and ship.

models, one key

paid, ever

OpenAI-compatible

Getting started

Authentication

Every /v1/* endpoint expects a bearer token. Keys are prefixed with obsidianx- followed by a 32-character secret. Keep it server-side — never ship it in a browser bundle.

request header

Authorization: Bearer obsidianx-<32-char-key>

Missing or bad key?

You'll get a 401 back immediately, before any tokens are billed. Rotate compromised keys right away.

Don't have a key?

Join our Discord and run /register to get your token API key instantly. It's the only place keys and credits come from — no sign-ups, no payments.

Join Discord

Quickstart

Already on the OpenAI SDK? Change two lines — the base URL and the key — and every model on the pricing table is yours.

curl https://obsidian-x-api.vercel.app/v1/chat/completions \
  -H "Authorization: Bearer $OBSIDIANX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek-v4-flash",
    "messages": [{ "role": "user", "content": "Hello" }]
  }'

from openai import OpenAI

client = OpenAI(
    base_url="https://obsidian-x-api.vercel.app/v1",
    api_key="obsidianx-...",
)

resp = client.chat.completions.create(
    model="deepseek-v4-flash",
    messages=[{"role": "user", "content": "Hello"}],
)
print(resp.choices[0].message.content)

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://obsidian-x-api.vercel.app/v1",
  apiKey: process.env.OBSIDIANX_API_KEY,
});

const resp = await client.chat.completions.create({
  model: "deepseek-v4-flash",
  messages: [{ role: "user", content: "Hello" }],
});
console.log(resp.choices[0].message.content);

POST /v1/chat/completions

OpenAI-compatible chat completion. Supports streaming via SSE, tool calling, JSON mode, deterministic seeds and reasoning tokens. Streaming responses are billed on the final aggregated usage.

Request body

request · JSON

{
  "model": "deepseek-v4-flash",
  "messages": [{ "role": "user", "content": "Hello" }],
  "stream": false,
  "temperature": 0.7,
  "max_tokens": 2048
}

response includes x_obsidian metadata 200

{
  "id": "chatcmpl-…", "choices": [ … ],
  "x_obsidian": {
    "cost": 0.0002,   // billed for this call ($)
    "balance": 9.54   // remaining balance ($)
  }
}

See how the model thinks

{ "object": "list", "data": [ { "id": "deepseek-v4-flash", "object": "model", "owned_by": "obsidian", "pricing": { "input_per_1k": 0.0008, "output_per_1k": 0.0012 } } ] }

One API.
Forty-seven models.
Zero rewrites.

Authentication

Quickstart

See how the model thinks

Models & pricing

Cost calculation

Error codes

Swap the base URL. That's the migration.

One API. Forty-seven models. Zero rewrites.

Authentication

Quickstart

See how the model thinks

Models & pricing

Cost calculation

Error codes

Swap the base URL. That's the migration.

One API.
Forty-seven models.
Zero rewrites.