diff --git a/packages/inference/src/tasks/custom/request.ts b/packages/inference/src/tasks/custom/request.ts
index c794ee2903..d8baa70626 100644
--- a/packages/inference/src/tasks/custom/request.ts
+++ b/packages/inference/src/tasks/custom/request.ts
@@ -16,7 +16,7 @@ export async function request<T>(
 	const { url, info } = await makeRequestOptions(args, options);
 	const response = await (options?.fetch ?? fetch)(url, info);
 
-	if (options?.retry_on_error !== false && response.status === 503 && !options?.wait_for_model) {
+	if (options?.retry_on_error !== false && response.status === 503 && options?.wait_for_model) {
 		return request(args, {
 			...options,
 			wait_for_model: true,
diff --git a/packages/inference/src/tasks/custom/streamingRequest.ts b/packages/inference/src/tasks/custom/streamingRequest.ts
index 07ec6b8d8a..c0c6cb0b3c 100644
--- a/packages/inference/src/tasks/custom/streamingRequest.ts
+++ b/packages/inference/src/tasks/custom/streamingRequest.ts
@@ -18,7 +18,7 @@ export async function* streamingRequest<T>(
 	const { url, info } = await makeRequestOptions({ ...args, stream: true }, options);
 	const response = await (options?.fetch ?? fetch)(url, info);
 
-	if (options?.retry_on_error !== false && response.status === 503 && !options?.wait_for_model) {
+	if (options?.retry_on_error !== false && response.status === 503 && options?.wait_for_model) {
 		return yield* streamingRequest(args, {
 			...options,
 			wait_for_model: true,