What causes "Failed to load inference settings (${response.status})." in different-ai/openwork?

Note the status and match it in Den server logs for GET /v1/inference; resolve the server-side cause.

different-ai/openwork · error · Error

Failed to load inference settings (${response.status}).

Error message

Failed to load inference settings (${response.status}).

What it means

loadStatus in InferenceScreen GETs /v1/inference (12s timeout) to load inference configuration. A non-OK response throws `Failed to load inference settings (${response.status})` unless the payload provides a message via getErrorMessage. Even on 2xx, if parseInferencePayload can't extract the settings, a separate "Inference settings response was incomplete." error is thrown. Errors are shown via setError in the UI.

Source

Thrown at ee/apps/den-web/app/(den)/dashboard/_components/inference-screen.tsx:322

  );
  const canManageModels = access.isAdmin;
  // OpenWork Models are a hosted OpenWork Cloud offering; self-hosted
  // (single-org) deployments manage their own LLM providers instead.
  const isSelfHosted = runtimeConfigLoaded && runtimeConfig.orgMode === "single_org";
  const activeOrgSlug = activeOrg?.slug ?? null;

  useEffect(() => {
    if (!isSelfHosted) return;
    router.replace(getCustomLlmProvidersRoute(activeOrgSlug));
  }, [isSelfHosted, activeOrgSlug, router]);

  async function loadStatus() {
    setLoading(true);
    setError(null);
    try {
      const { response, payload } = await requestJson("/v1/inference", { method: "GET" }, 12000);
      if (!response.ok) {
        throw new Error(getErrorMessage(payload, `Failed to load inference settings (${response.status}).`));
      }
      const parsed = parseInferencePayload(payload);
      if (!parsed) {
        throw new Error("Inference settings response was incomplete.");
      }
      setStatus(parsed);
    } catch (loadError) {
      setError(loadError instanceof Error ? loadError.message : "Failed to load inference settings.");
    } finally {
      setLoading(false);
    }
  }

  useEffect(() => {
    void loadStatus();
  }, [orgContext?.organization.id]);

  // Subscribe at the point of value: start the Stripe checkout right here

View on GitHub (pinned to 2b7df46e8a)

Solutions

Note the status and match it in Den server logs for GET /v1/inference; resolve the server-side cause.
On 401, sign out/in to refresh the Den session and reload the screen.
On 403, obtain the org role required for viewing/changing inference settings.
On 404, upgrade the self-hosted Den server to a version with /v1/inference.
For 5xx or timeouts, check Den service health/connectivity and retry.

Example fix

// before
const { response, payload } = await requestJson("/v1/inference", { method: "GET" }, 12000);
// expired session -> 401 -> "Failed to load inference settings (401)."

// after: refresh auth before loading
await refreshDenSession();
await loadStatus(); // 200 -> parseInferencePayload succeeds

Defensive patterns

Strategy: try-catch

Validate before calling

if (!denSessionActive()) await refreshSession(); // avoid predictable 401 before loadStatus()

Type guard

function isHttpError(e: unknown): e is Error & { status?: number } {
  return e instanceof Error && /\(\d{3}\)/.test(e.message);
}

Try / catch

try {
  await loadStatus();
} catch (e) {
  if (isHttpError(e) && e.message.includes("401")) {
    promptReSignIn();
  } else {
    showError(e instanceof Error ? e.message : "Failed to load inference settings.");
  }
}

Prevention

Refresh the Den session before data loads in long-lived dashboards.
Check role permissions before showing the inference screen.
Keep self-hosted Den current so /v1/inference exists.
Retry once with backoff on transient 5xx or the 12s timeout.

When it happens

Trigger: GET /v1/inference returns 401 (expired session), 403 (insufficient permissions), 404 (Den version without the endpoint), 5xx (backend error), or times out at 12s. Distinct from the payload-shape failure, which yields the "incomplete" message instead.

Common situations: Idle dashboard with expired Den session; member-role user opening inference settings that need admin; older self-hosted Den lacking /v1/inference; Den API briefly unavailable during deploy or restart.

Understand the failure class

Background: 'Something went wrong' / 'Request failed (500)' / 'HTTP error! status: 404' — what failed HTTP requests actually mean and how to find the real cause — this error's family across 28 libraries.

Related errors

AI-assisted analysis of different-ai/openwork@2b7df46e8a (2026-09-01). Data as JSON: /api/errors/c1235b15c9a73d78. Report an issue: GitHub.