diff --git a/index.d.ts b/index.d.ts
index bf9cd5d..c415bf0 100644
--- a/index.d.ts
+++ b/index.d.ts
@@ -164,6 +164,18 @@ declare module 'replicate' {
           version_id: string
         ): Promise<ModelVersion>;
       };
+      predictions: {
+        create(
+          model_owner: string,
+          model_name: string,
+          options: {
+            input: object;
+            stream?: boolean;
+            webhook?: string;
+            webhook_events_filter?: WebhookEventType[];
+          }
+        ): Promise<Prediction>;
+      };
     };
 
     predictions: {
diff --git a/index.js b/index.js
index acb07eb..0552ea5 100644
--- a/index.js
+++ b/index.js
@@ -68,6 +68,9 @@ class Replicate {
         list: models.versions.list.bind(this),
         get: models.versions.get.bind(this),
       },
+      predictions: {
+        create: models.predictions.create.bind(this),
+      },
     };
 
     this.predictions = {
diff --git a/index.test.ts b/index.test.ts
index d033902..ec9e523 100644
--- a/index.test.ts
+++ b/index.test.ts
@@ -668,6 +668,38 @@ describe('Replicate client', () => {
     // Add more tests for error handling, edge cases, etc.
   });
 
+  describe('models.predictions.create', () => {
+    test('Calls the correct API route with the correct payload', async () => {
+      nock(BASE_URL)
+        .post('/models/meta/llama-2-70b-chat/predictions')
+        .reply(200, {
+          id: "heat2o3bzn3ahtr6bjfftvbaci",
+          model: "replicate/lifeboat-70b",
+          version: "d-c6559c5791b50af57b69f4a73f8e021c",
+          input: {
+            prompt: "Please write a haiku about llamas."
+          },
+          logs: "",
+          error: null,
+          status: "starting",
+          created_at: "2023-11-27T13:35:45.99397566Z",
+          urls: {
+            cancel: "https://api.replicate.com/v1/predictions/heat2o3bzn3ahtr6bjfftvbaci/cancel",
+            get: "https://api.replicate.com/v1/predictions/heat2o3bzn3ahtr6bjfftvbaci"
+          }
+        });
+      const prediction = await client.models.predictions.create("meta", "llama-2-70b-chat", {
+        input: {
+          prompt: "Please write a haiku about llamas."
+        },
+        webhook: 'http://test.host/webhook',
+        webhook_events_filter: [ 'output', 'completed' ],
+      });
+      expect(prediction.id).toBe('heat2o3bzn3ahtr6bjfftvbaci');
+    });
+    // Add more tests for error handling, edge cases, etc.
+  });
+
   describe('hardware.list', () => {
     test('Calls the correct API route', async () => {
       nock(BASE_URL)
diff --git a/lib/models.js b/lib/models.js
index 3c4e5b1..12419fc 100644
--- a/lib/models.js
+++ b/lib/models.js
@@ -83,9 +83,33 @@ async function createModel(model_owner, model_name, options) {
   return response.json();
 }
 
+/**
+ * Create a new prediction
+ *
+ * @param {string} model_owner - Required. The name of the user or organization that owns the model
+ * @param {string} model_name - Required. The name of the model
+ * @param {object} options
+ * @param {object} options.input - Required. An object with the model inputs
+ * @param {string} [options.webhook] - An HTTPS URL for receiving a webhook when the prediction has new output
+ * @param {string[]} [options.webhook_events_filter] - You can change which events trigger webhook requests by specifying webhook events (`start`|`output`|`logs`|`completed`)
+ * @param {boolean} [options.stream] - Whether to stream the prediction output. Defaults to false
+ * @returns {Promise<object>} Resolves with the created prediction
+ */
+async function createPrediction(model_owner, model_name, options) {
+  const { stream, ...data } = options;
+
+  const response = await this.request(`/models/${model_owner}/${model_name}/predictions`, {
+    method: 'POST',
+    data: { ...data, stream },
+  });
+
+  return response.json();
+}
+
 module.exports = {
   get: getModel,
   list: listModels,
   create: createModel,
   versions: { list: listModelVersions, get: getModelVersion },
+  predictions: { create: createPrediction },
 };