Skip to content

Never stop coding. Free MIT AI gateway: one endpoint, 375 providers (150+ free), 1,400+ models Kimi, Claude, GPT, Gemini, GLM, DeepSeek, MiniMax. Works with Claude Code, Codex, Cursor, OpenCode, Cline & Copilot. Quota-aware auto-fallback, RTK+Caveman compression saves 15-95% tokens, MCP/A2A, Desktop/PWA. Built by 900+ contributors

4.6 KB
Raw Blame
import { NextResponse } from "next/server";
import { z } from "zod";
import { requireManagementAuth } from "@/lib/api/requireManagementAuth";
import { DEFAULT_MODEL_TEST_TIMEOUT_MS, runSingleModelTest } from "@/lib/api/modelTestRunner";
import { getActiveSyncedCatalog } from "@/lib/db/models/activeSyncedCatalog";
import { hasNoRunnableEndpointAdapter } from "@/lib/providerModels/endpointAdapterAvailability";
import { isModelEndpointCapabilityGateEnabled } from "@/lib/providerModels/endpointCapabilityGateFlag";
import { sanitizeErrorMessage } from "@omniroute/open-sse/utils/error.ts";
import { getSettings } from "@/lib/db/settings";
import { isFreeModel, providerHasFreeModels } from "@/shared/utils/freeModels";

const testModelSchema = z.object({
  providerId: z.string().min(1),
  modelId: z.string().min(1),
  connectionId: z.string().min(1).optional(),
});

const NVIDIA_SINGLE_TEST_TIMEOUT_MS = 180_000;
export async function POST(request: Request) {
  const authError = await requireManagementAuth(request);
  if (authError) return authError;

  let rawBody;
  try {
    rawBody = await request.json();
  } catch {
    // Keep `error` a plain string — the dashboard renders it directly in a toast,
    // and an object here throws React #31 ("Objects are not valid as a React
    // child"), freezing the whole page instead of showing the message.
    return NextResponse.json({ status: "error", error: "Invalid JSON body" }, { status: 400 });
  }

  try {
    const validation = testModelSchema.safeParse(rawBody);
    if (!validation.success) {
      // Flatten the Zod issues to a string (never return the object — see above).
      const detail = validation.error.issues
        .map((i) => `${i.path.join(".") || "body"}: ${i.message}`)
        .join("; ");
      return NextResponse.json(
        { status: "error", error: `Invalid request: ${detail}` },
        { status: 400 }
      );
    }
    const { providerId, modelId, connectionId } = validation.data;
    // #16929 (opt-in): a trusted endpoint tag with no provider adapter is answered
    // locally with 422 instead of dispatching a doomed test. The flag is read first so
    // the default path does no extra catalog read.
    if (isModelEndpointCapabilityGateEnabled()) {
      let unrunnable = false;
      try {
        const catalog = await getActiveSyncedCatalog(providerId);
        const model = catalog.models.find((row) => row.id === modelId);
        unrunnable = hasNoRunnableEndpointAdapter(providerId, model);
      } catch {
        // Fail open: a catalog read miss keeps the normal test dispatch.
      }
      if (unrunnable) {
        return NextResponse.json(
          { status: "error", error: "Provider has no runnable adapter for this model's endpoint." },
          { status: 422 }
        );
      }
    }

    // #6328 (follow-up to #6495): REMOVE — not just hide — paid Test dispatches
    // when hidePaidModels is on. 403 (distinct from validation-400) so callers
    // can tell policy-blocked apart from bad-request. Fail open on settings read.
    let hidePaid = false;
    try {
      const settings = await getSettings();
      hidePaid = settings?.hidePaidModels === true;
    } catch {}
    if (
      hidePaid &&
      !(providerHasFreeModels(providerId) && isFreeModel(providerId, { id: modelId }))
    ) {
      return NextResponse.json(
        {
          status: "error",
          error: "Paid model blocked while hidePaidModels is enabled",
        },
        { status: 403 }
      );
    }

    const result = await runSingleModelTest({
      providerId,
      modelId,
      ...(connectionId ? { connectionId } : {}),
      timeoutMs:
        providerId.trim().toLowerCase() === "nvidia"
          ? NVIDIA_SINGLE_TEST_TIMEOUT_MS
          : DEFAULT_MODEL_TEST_TIMEOUT_MS,
      streamChat: true,
    });

    if (result.status === "ok") {
      return NextResponse.json({
        status: "ok",
        latencyMs: result.latencyMs,
        responseText: result.responseText,
      });
    }

    const body: Record<string, unknown> = {
      status: "error",
      latencyMs: result.latencyMs,
      error: result.error || "Unknown error",
    };
    if (result.statusCode !== undefined) body.statusCode = result.statusCode;
    if (result.rateLimited) body.rateLimited = true;
    if (result.retryAfter !== undefined) body.retryAfter = result.retryAfter;
    if (result.skipped) body.skipped = true;

    return NextResponse.json(body, { status: result.httpStatus });
  } catch (error: unknown) {
    return NextResponse.json(
      {
        status: "error",
        error: sanitizeErrorMessage(error) || "Unknown error",
      },
      { status: 500 }
    );
  }
}