API reference

GET /models

List the curated model ids.

Krallo exposes a small, fixed catalogue of curated model ids. Each id is a stable public alias — the underlying provider and model version can change without breaking your integration, and Krallo never exposes the internal routing string a public id maps to. Use this endpoint to discover the valid ids, then pass one to the model field of Chat completions.

The response is deliberately OpenAI-shaped ({ "object": "list", "data": [ … ] }) so that OpenAI-compatible clients — opencode, Aider, Continue, nanocoder, or any SDK pointed at base URL https://exe.krallo.com/api/v1 — can probe it during startup and populate a model picker automatically.

Request

Authenticated with a bearer key. This endpoint requires a valid key only — it does not require any specific scope, unlike /chat/completions (which needs ai:invoke). It is rate-limited per key but not metered: listing models never draws down your credit balance.

See Authentication for how to send the key.

Parameters

None. The endpoint takes no path, query, or body parameters. Any query string is ignored, and a request body is not read.

Example request

curl https://exe.krallo.com/api/v1/models \
  -H "Authorization: Bearer $KRALLO_API_KEY"

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.KRALLO_API_KEY,
  baseURL: "https://exe.krallo.com/api/v1",
});

const models = await client.models.list();
for (const m of models.data) {
  console.log(m.id); // krallo-chat, krallo-code, krallo-fast, krallo-reasoning
}

The curated model ids

Four public ids are available. All are owned_by krallo. Pick by the shape of the work, not by a specific underlying model — Krallo may re-point an alias to a newer or better model at any time.

Response

200 OK with an OpenAI-shaped list envelope. Each element is a minimal model object — Krallo returns only id, object, and owned_by; it does not return context windows, pricing, or created timestamps.

Example response

{
  "object": "list",
  "data": [
    { "id": "krallo-chat",      "object": "model", "owned_by": "krallo" },
    { "id": "krallo-code",      "object": "model", "owned_by": "krallo" },
    { "id": "krallo-fast",      "object": "model", "owned_by": "krallo" },
    { "id": "krallo-reasoning", "object": "model", "owned_by": "krallo" }
  ]
}

Every authenticated response also carries advisory rate-limit headers — X-RateLimit-Limit, X-RateLimit-Remaining, and X-RateLimit-Reset — reflecting your per-key budget.

Passing a model to chat completions

The id values from this list are exactly what Chat completions accepts in its model field:

{
  "model": "krallo-reasoning",
  "messages": [
    { "role": "user", "content": "Prove that the sum of two even integers is even." }
  ]
}

• Omitting or leaving model blank falls back to krallo-code.
• An unrecognized id is rejected by the completions endpoint with 400 model_not_found; the error message lists the valid ids. Calling GET /models first avoids that round-trip.

Errors

The catalogue is fixed, so this endpoint has no request-validation errors of its own beyond the wrong HTTP method — the remaining failures come from the shared auth and rate-limit gates.