fix: list Fantastico models from the control plane and add its icon

The model catalog lives on the control plane, not on an OpenAI-style
/models route, so autodetect hit a 404 on /api/models. listModels now
reads /edge/models for account keys and /runtime/models for project keys.

Also adds the Fantastico logo to the provider picker.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

Author: lucabyte42-lab

core/llm/llms/Fantastico.ts

@@ -16,6 +16,21 @@ class Fantastico extends OpenAI {
   };
 
   protected useOpenAIAdapterFor: (LlmApiRequestType | "*")[] = ["*"];
+
+  /**
+   * The catalog is served by the control plane, not by the inference endpoint:
+   * account keys read it from `/edge/models`, project keys from `/runtime/models`.
+   */
+  async listModels(): Promise<string[]> {
+    const route = this.apiKey?.startsWith("sk-fant-") ? "runtime" : "edge";
+    const controlPlane = (this.apiBase ?? "").replace(/\/+$/, "");
+    const response = await this.fetch(`${controlPlane}/${route}/models`, {
+      method: "GET",
+      headers: { Authorization: `Bearer ${this.apiKey}` },
+    });
+    const data = (await response.json()) as { models?: { id: string }[] };
+    return (data.models ?? []).map((m) => m.id);
+  }
 }
 
 export default Fantastico;

core/llm/llms/Fantastico.vitest.ts

@@ -0,0 +1,59 @@
+import { describe, expect, it, vi } from "vitest";
+
+import Fantastico from "./Fantastico";
+
+function mockModels(fantastico: Fantastico) {
+  return vi.spyOn(fantastico, "fetch").mockResolvedValue(
+    new Response(
+      JSON.stringify({
+        models: [{ id: "anthropic/claude-sonnet-4.5" }, { id: "openai/gpt-5" }],
+      }),
+    ),
+  );
+}
+
+describe("Fantastico", () => {
+  it("should have correct provider name", () => {
+    expect(Fantastico.providerName).toBe("fantastico");
+  });
+
+  it("should default to the production control plane", () => {
+    expect(Fantastico.defaultOptions.apiBase).toBe(
+      "https://fantastico.cc/api/",
+    );
+  });
+
+  it("should list models from /edge/models with an account key", async () => {
+    const fantastico = new Fantastico({
+      model: "AUTODETECT",
+      apiKey: "ck-fant-test",
+    });
+    const fetchSpy = mockModels(fantastico);
+
+    const models = await fantastico.listModels();
+
+    expect(models).toEqual(["anthropic/claude-sonnet-4.5", "openai/gpt-5"]);
+    expect(fetchSpy).toHaveBeenCalledWith(
+      "https://fantastico.cc/api/edge/models",
+      expect.objectContaining({
+        headers: { Authorization: "Bearer ck-fant-test" },
+      }),
+    );
+  });
+
+  it("should list models from /runtime/models with a project key", async () => {
+    const fantastico = new Fantastico({
+      model: "AUTODETECT",
+      apiKey: "sk-fant-test",
+      apiBase: "https://fantastico.example/api",
+    });
+    const fetchSpy = mockModels(fantastico);
+
+    await fantastico.listModels();
+
+    expect(fetchSpy).toHaveBeenCalledWith(
+      "https://fantastico.example/api/runtime/models",
+      expect.anything(),
+    );
+  });
+});

gui/public/logos/fantastico.svg

@@ -1273,6 +1273,7 @@ To get started, [register](https://dataplatform.cloud.ibm.com/registration/stepo
   fantastico: {
     title: "Fantastico",
     provider: "fantastico",
+    icon: "fantastico.svg",
     description:
       "Per-developer account keys with preallocated budgets and one company bill. No proxy: requests go straight to the model provider.",
     longDescription: