diff --git a/src/backends/onnx.js b/src/backends/onnx.js
index a64f9d160..b7d4ec7bc 100644
--- a/src/backends/onnx.js
+++ b/src/backends/onnx.js
@@ -56,12 +56,15 @@ let defaultDevices;
 let ONNX;
 const ORT_SYMBOL = Symbol.for('onnxruntime');
 
+/** @type {"custom"|"node"|"web"} */
+let ort;
 if (ORT_SYMBOL in globalThis) {
     // If the JS runtime exposes their own ONNX runtime, use it
     ONNX = globalThis[ORT_SYMBOL];
+    ort = 'custom';
 
-} else if (apis.IS_NODE_ENV) {
-    ONNX = ONNX_NODE.default ?? ONNX_NODE;
+} else if (apis.IS_NODE_ENV && (ONNX = ONNX_NODE.default ?? ONNX_NODE)?.InferenceSession) {
+    ort = 'node';
 
     // Updated as of ONNX Runtime 1.20.1
     // The following table lists the supported versions of ONNX Runtime Node.js binding provided with pre-built binaries.
@@ -87,6 +90,7 @@ if (ORT_SYMBOL in globalThis) {
     defaultDevices = ['cpu'];
 } else {
     ONNX = ONNX_WEB;
+    ort = 'web';
 
     if (apis.IS_WEBNN_AVAILABLE) {
         // TODO: Only push supported providers (depending on available hardware)
@@ -169,6 +173,14 @@ export function isONNXTensor(x) {
     return x instanceof ONNX.Tensor;
 }
 
+/**
+ * The type of ONNX runtime being used.
+ * - 'node' for `onnxruntime-node`
+ * - 'web' for `onnxruntime-web`
+ * - 'custom' for a custom ONNX runtime
+ */
+export const runtime = ort;
+
 /** @type {import('onnxruntime-common').Env} */
 // @ts-ignore
 const ONNX_ENV = ONNX?.env;
diff --git a/src/env.js b/src/env.js
index fd3924213..39693edd5 100644
--- a/src/env.js
+++ b/src/env.js
@@ -142,7 +142,7 @@ export const env = {
     remoteHost: 'https://huggingface.co/',
     remotePathTemplate: '{model}/resolve/{revision}/',
 
-    allowLocalModels: !(IS_BROWSER_ENV || IS_WEBWORKER_ENV),
+    allowLocalModels: !(IS_BROWSER_ENV || IS_WEBWORKER_ENV || IS_DENO_RUNTIME),
     localModelPath: localModelPath,
     useFS: IS_FS_AVAILABLE,
 
diff --git a/src/models.js b/src/models.js
index f05a58665..cff3ee8b6 100644
--- a/src/models.js
+++ b/src/models.js
@@ -48,6 +48,7 @@ import {
     createInferenceSession,
     isONNXTensor,
     isONNXProxy,
+    runtime,
 } from './backends/onnx.js';
 import {
     DATA_TYPES,
@@ -172,7 +173,7 @@ async function getSession(pretrained_model_name_or_path, fileName, options) {
 
     // If the device is not specified, we use the default (supported) execution providers.
     const selectedDevice = /** @type {import("./utils/devices.js").DeviceType} */(
-        device ?? (apis.IS_NODE_ENV ? 'cpu' : 'wasm')
+        device ?? (runtime === "web" ? 'wasm' : 'cpu')
     );
 
     const executionProviders = deviceToExecutionProviders(selectedDevice);
diff --git a/webpack.config.js b/webpack.config.js
index 53b5ccd9d..52d2c8c1d 100644
--- a/webpack.config.js
+++ b/webpack.config.js
@@ -54,7 +54,21 @@ class PostBuildPlugin {
       {
         const src = path.join(__dirname, 'node_modules/onnxruntime-web/dist', ORT_JSEP_FILE);
         const dest = path.join(dist, ORT_JSEP_FILE);
-        fs.copyFileSync(src, dest);
+
+        // Transformers.js uses both onnxruntime-web and onnxruntime-node in the same package,
+        // and the runtime we use depends on the environment (onnxruntime-web for web, onnxruntime-node for Node.js).
+        // This means that we don't currently support using the WASM backend in Node.js, so we disable this behaviour in the JSEP file.
+        const content = fs.readFileSync(src, 'utf8');
+        const updatedContent = content
+          .replace(
+            `"object"==typeof process&&"object"==typeof process.versions&&"string"==typeof process.versions.node&&"renderer"!=process.type`,
+            "false",
+          )
+          .replace(
+            `typeof globalThis.process?.versions?.node == 'string'`,
+            "false",
+          )
+        fs.writeFileSync(dest, updatedContent);
       }
     });
   }