cencori 1.3.0 → 1.3.1

package/README.md

@@ -130,6 +130,46 @@ const response = await cencori.ai.embeddings({
 console.log(response.embeddings[0]); // [0.1, 0.2, ...]
 ```
 
+### Vision (Image Understanding)
+
+Analyze, describe, OCR, and classify images. Routes across OpenAI, Anthropic, and Google vision-capable models.
+
+```typescript
+// General analysis with a custom prompt
+const result = await cencori.vision.analyze({
+  image: { url: 'https://example.com/photo.jpg' },
+  prompt: 'What breed of dog is this?'
+});
+
+// OCR from a local file
+import fs from 'node:fs';
+const buf = fs.readFileSync('receipt.png');
+const { text } = await cencori.vision.ocr({
+  image: { base64: buf.toString('base64'), mimeType: 'image/png' }
+});
+
+// Structured classification
+const { classification } = await cencori.vision.classify({
+  image: { url: 'https://example.com/product.jpg' }
+});
+// classification: { primary_category, tags, objects, safe_for_work, ... }
+```
+
+Drop-in React uploader for your own product (`cencori/react`):
+
+```tsx
+import { VisionUploader } from 'cencori/react';
+
+<VisionUploader
+  endpoint="https://cencori.com/api/ai/vision"
+  apiKey={process.env.NEXT_PUBLIC_CENCORI_KEY}
+  task="describe"
+  onResult={(r) => console.log(r)}
+/>
+```
+
+Full API in [docs](https://cencori.com/docs/ai/endpoints/vision).
+
 ### Image Generation
 
 Generate images from text prompts using multiple providers:

package/dist/index.d.mts

@@ -3,7 +3,7 @@ export { j as ChatMessage, a as ChatRequest, b as ChatResponse, o as CodeInterpr
 import { AINamespace } from './ai/index.mjs';
 export { StreamChunk } from './ai/index.mjs';
 import { VisionNamespace } from './vision/index.mjs';
-export { VisionClassification, VisionClassifyResult, VisionCost, VisionDescribeResult, VisionImage, VisionOcrResult, VisionProvider, VisionRequest, VisionResult, VisionTask, VisionUsage } from './vision/index.mjs';
+export { VisionClassification, VisionClassifyResult, VisionCost, VisionDescribeResult, VisionImage, VisionOcrResult, VisionProvider, VisionRequest, VisionResult, VisionStreamChunk, VisionTask, VisionUsage } from './vision/index.mjs';
 import { ComputeNamespace } from './compute/index.mjs';
 import { WorkflowNamespace } from './workflow/index.mjs';
 import { StorageNamespace } from './storage/index.mjs';

package/dist/index.d.ts

@@ -3,7 +3,7 @@ export { j as ChatMessage, a as ChatRequest, b as ChatResponse, o as CodeInterpr
 import { AINamespace } from './ai/index.js';
 export { StreamChunk } from './ai/index.js';
 import { VisionNamespace } from './vision/index.js';
-export { VisionClassification, VisionClassifyResult, VisionCost, VisionDescribeResult, VisionImage, VisionOcrResult, VisionProvider, VisionRequest, VisionResult, VisionTask, VisionUsage } from './vision/index.js';
+export { VisionClassification, VisionClassifyResult, VisionCost, VisionDescribeResult, VisionImage, VisionOcrResult, VisionProvider, VisionRequest, VisionResult, VisionStreamChunk, VisionTask, VisionUsage } from './vision/index.js';
 import { ComputeNamespace } from './compute/index.js';
 import { WorkflowNamespace } from './workflow/index.js';
 import { StorageNamespace } from './storage/index.js';

package/dist/index.js

@@ -676,6 +676,57 @@ var VisionNamespace = class {
   async classify(request) {
     return this.post("/api/ai/vision/classify", request);
   }
+  /**
+   * Stream a vision analysis. Yields `{ delta }` chunks as text arrives, then
+   * a final `{ done: true, model, provider, usage, cost }` chunk with totals.
+   *
+   * @example
+   * for await (const chunk of cencori.vision.analyzeStream({ image: { url } })) {
+   *   if (chunk.delta) process.stdout.write(chunk.delta);
+   *   if (chunk.done) console.log(chunk.cost);
+   * }
+   */
+  async *analyzeStream(request) {
+    const body = { ...toBody(request), stream: true };
+    const response = await fetch(`${this.config.baseUrl}/api/ai/vision`, {
+      method: "POST",
+      headers: {
+        "CENCORI_API_KEY": this.config.apiKey,
+        "Content-Type": "application/json",
+        ...this.config.headers
+      },
+      body: JSON.stringify(body)
+    });
+    if (!response.ok) {
+      const data = await response.json().catch(() => ({}));
+      const message = data && typeof data === "object" && "message" in data && typeof data.message === "string" ? data.message : `Vision stream request failed with status ${response.status}`;
+      throw new Error(message);
+    }
+    if (!response.body) throw new Error("Response body is null");
+    const reader = response.body.getReader();
+    const decoder = new TextDecoder();
+    let buffer = "";
+    try {
+      while (true) {
+        const { done, value } = await reader.read();
+        if (done) break;
+        buffer += decoder.decode(value, { stream: true });
+        const lines = buffer.split("\n");
+        buffer = lines.pop() ?? "";
+        for (const line of lines) {
+          if (!line.startsWith("data: ")) continue;
+          const data = line.slice(6);
+          if (data === "[DONE]") return;
+          try {
+            yield JSON.parse(data);
+          } catch {
+          }
+        }
+      }
+    } finally {
+      reader.releaseLock();
+    }
+  }
   async post(path, request) {
     const response = await fetch(`${this.config.baseUrl}${path}`, {
       method: "POST",