@canarycoders/ai 0.8.0 → 0.8.1

package/dist/index.d.cts

@@ -6,7 +6,7 @@ export { b as RealtimeKind, c as RealtimeTool } from './realtime-BWDkXcj9.cjs';
 type FetchLike = typeof globalThis.fetch;
 
 /** Every provider the gateway can route to. */
-type ProviderType = "gemini" | "vertex" | "openai" | "anthropic" | "xai" | "lmstudio" | "ollama" | "perplexity" | "elevenlabs" | "mlxaudio" | "vision" | "gcvision" | "zhipu" | "moonshot";
+type ProviderType = "gemini" | "vertex" | "openai" | "anthropic" | "xai" | "lmstudio" | "ollama" | "perplexity" | "elevenlabs" | "mlxaudio" | "vision" | "gcvision" | "zhipu" | "moonshot" | "ocr" | "ltx";
 /** Providers that serve text chat / completion. */
 type ChatProvider = "gemini" | "vertex" | "openai" | "anthropic" | "xai" | "lmstudio" | "ollama" | "perplexity" | "zhipu" | "moonshot";
 type ModelCapability = "chat" | "vision" | "image-generation" | "video-generation" | "text-to-speech" | "speech-to-text" | "image-recognition" | "face-detection" | "web-detection" | "conversation" | "sound-effect" | "music-generation" | "dialogue" | "realtime-voice" | "realtime-translate" | "realtime-transcribe" | "embeddings" | "reasoning";
@@ -509,15 +509,26 @@ interface ImageGenerationResult {
     cost?: CostInfo;
 }
 interface VideoGenerateParams {
-    provider: "gemini" | "vertex" | "xai";
+    /** gemini/vertex = Veo 3.1, xai = Grok Imagine, ltx = local LTX-2 (no provider cost). */
+    provider: "gemini" | "vertex" | "xai" | "ltx";
     prompt: string;
     model?: string;
+    /**
+     * gemini/vertex/xai: "16:9" or "9:16". ltx: any "W:H"; picks a 64-grid size
+     * (16:9 = 768x448, 9:16 = 448x768) unless `resolution` is set.
+     */
     aspectRatio?: string;
+    /**
+     * gemini/vertex: "720p" | "1080p". xai: "480p" | "720p". ltx: pixels "WxH"
+     * (e.g. "768x512"), each side snapped to a multiple of 64 (256..1536).
+     */
     resolution?: string;
     durationSeconds?: number;
     numberOfVideos?: number;
-    /** public URL of a seed image */
+    /** public URL (or data URI) of the start frame for image-to-video */
     imageUrl?: string;
+    /** Veo only: final frame, interpolates first to last. Rejected by ltx. */
+    lastFrameUrl?: string;
     /** id from `video.upload()` for image-to-video */
     fileId?: string;
     tag?: string;

package/dist/index.d.ts

@@ -6,7 +6,7 @@ export { b as RealtimeKind, c as RealtimeTool } from './realtime-BWDkXcj9.js';
 type FetchLike = typeof globalThis.fetch;
 
 /** Every provider the gateway can route to. */
-type ProviderType = "gemini" | "vertex" | "openai" | "anthropic" | "xai" | "lmstudio" | "ollama" | "perplexity" | "elevenlabs" | "mlxaudio" | "vision" | "gcvision" | "zhipu" | "moonshot";
+type ProviderType = "gemini" | "vertex" | "openai" | "anthropic" | "xai" | "lmstudio" | "ollama" | "perplexity" | "elevenlabs" | "mlxaudio" | "vision" | "gcvision" | "zhipu" | "moonshot" | "ocr" | "ltx";
 /** Providers that serve text chat / completion. */
 type ChatProvider = "gemini" | "vertex" | "openai" | "anthropic" | "xai" | "lmstudio" | "ollama" | "perplexity" | "zhipu" | "moonshot";
 type ModelCapability = "chat" | "vision" | "image-generation" | "video-generation" | "text-to-speech" | "speech-to-text" | "image-recognition" | "face-detection" | "web-detection" | "conversation" | "sound-effect" | "music-generation" | "dialogue" | "realtime-voice" | "realtime-translate" | "realtime-transcribe" | "embeddings" | "reasoning";
@@ -509,15 +509,26 @@ interface ImageGenerationResult {
     cost?: CostInfo;
 }
 interface VideoGenerateParams {
-    provider: "gemini" | "vertex" | "xai";
+    /** gemini/vertex = Veo 3.1, xai = Grok Imagine, ltx = local LTX-2 (no provider cost). */
+    provider: "gemini" | "vertex" | "xai" | "ltx";
     prompt: string;
     model?: string;
+    /**
+     * gemini/vertex/xai: "16:9" or "9:16". ltx: any "W:H"; picks a 64-grid size
+     * (16:9 = 768x448, 9:16 = 448x768) unless `resolution` is set.
+     */
     aspectRatio?: string;
+    /**
+     * gemini/vertex: "720p" | "1080p". xai: "480p" | "720p". ltx: pixels "WxH"
+     * (e.g. "768x512"), each side snapped to a multiple of 64 (256..1536).
+     */
     resolution?: string;
     durationSeconds?: number;
     numberOfVideos?: number;
-    /** public URL of a seed image */
+    /** public URL (or data URI) of the start frame for image-to-video */
     imageUrl?: string;
+    /** Veo only: final frame, interpolates first to last. Rejected by ltx. */
+    lastFrameUrl?: string;
     /** id from `video.upload()` for image-to-video */
     fileId?: string;
     tag?: string;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@canarycoders/ai",
-  "version": "0.8.0",
+  "version": "0.8.1",
   "description": "Official TypeScript SDK for the CanaryCoders AI multi-provider gateway",
   "license": "MIT",
   "author": "CanaryCoders",