@shop-os/modelkit 0.10.0 → 0.10.2

package/CHANGELOG.md

@@ -9,6 +9,23 @@ header in the release PR.
 
 ## [Unreleased]
 
+## [0.10.2] - Unreleased
+
+Wan 3 Prime on fal.
+
+### Added
+- `wan-3.0-prime` (Wan 3 Prime) video row on the fal adapter (`alibaba/wan-3.0-prime`), listed on `cowork` and `space_builder`. fal's Prime endpoints take the same inputs as Wan 3.0, so the row copies `wan-3.0`'s constraints and the existing Wan mapper routes it to `/text-to-video`, `/image-to-video` or `/reference-to-video` from the inputs. It is a separate row because Prime bills more ($0.068/s at 480p, $0.14/s at 720p, $0.28/s at 1080p on fal); the `credit_pricing_video` rows ship in the consumer bump.
+
+## [0.10.1] - Unreleased
+
+Fish Audio rows listed on every surface (ENG-3400).
+
+### Changed
+- `fish-s2.1-pro`, `fish-s2.1-pro-free`, `fish-drama-3`, `fish-voice-clone` and `fish-voice-design-1` move from `exposure: []` to `["cowork", "space_builder"]`. pro-backend prices the speech rows per 1,000 UTF-8 bytes of the script, the unit Fish bills; the seeds ship in the consumer bump.
+
+### Notes
+- The catalog and pricing seeds insert a row once, so environments that already hold a Fish row keep its old exposure; edit those rows in pro-admin. As of this release no environment holds them.
+
 ## [0.10.0] - Unreleased
 
 ElevenLabs direct: text to speech and instant voice cloning on our own ElevenLabs account (ENG-3400).

package/dist/catalog.cjs

@@ -2076,6 +2076,7 @@ var FISH_VOICES = [
 var FISH_LOGO = "https://cdn.jsdelivr.net/npm/@lobehub/icons-static-png@1.97.0/light/fishaudio.png";
 var PER_CALL2 = { unit: "images" };
 var PER_REQUEST2 = { unit: "requests" };
+var EXPOSURE4 = ["cowork", "space_builder"];
 var TTS_MAX_CHARS = 1e4;
 function fishTts(id, label, providerModelId, description) {
   return {
@@ -2086,7 +2087,7 @@ function fishTts(id, label, providerModelId, description) {
     adapter: "fish",
     providerModelId,
     lifecycle: "active",
-    exposure: [],
+    exposure: EXPOSURE4,
     constraints: {
       voices: FISH_VOICES,
       outputFormats: ["mp3", "wav"],
@@ -2127,7 +2128,7 @@ var FISH_MODELS = [
     adapter: "fish",
     providerModelId: "voice-clone",
     lifecycle: "active",
-    exposure: [],
+    exposure: EXPOSURE4,
     constraints: {
       // POST /model takes 1-20 voice files; Fish advises 10 s or more of clean speech each.
       references: { audios: 20, audioFormats: ["wav", "mp3", "m4a", "opus"] },
@@ -2151,7 +2152,7 @@ var FISH_MODELS = [
     adapter: "fish",
     providerModelId: "voice-design-1",
     lifecycle: "active",
-    exposure: [],
+    exposure: EXPOSURE4,
     constraints: {
       voiceDesign: { maxCandidates: 4, previewMaxChars: 150 },
       supports: [],
@@ -3749,7 +3750,7 @@ var GOOGLE_LOGO = "https://d31vvfpk2ppdyq.cloudfront.net/assets/image/192485d2-a
 var MINIMAX_LOGO = "https://d31vvfpk2ppdyq.cloudfront.net/assets/image/f4c2e906-28b7-4a4a-b491-df8fef7808ed.png";
 var SYNC_LOGO = "https://cdn.jsdelivr.net/npm/@lobehub/icons-static-png@1.97.0/light/sync.png";
 var PER_CALL3 = { unit: "images" };
-var EXPOSURE4 = ["cowork", "space_builder"];
+var EXPOSURE5 = ["cowork", "space_builder"];
 var LIPSYNC_AUDIO = { audioSeconds: { max: 30 } };
 function elevenLabsTts(tts) {
   return {
@@ -3760,7 +3761,7 @@ function elevenLabsTts(tts) {
     adapter: "fal",
     providerModelId: tts.providerModelId,
     lifecycle: "active",
-    exposure: EXPOSURE4,
+    exposure: EXPOSURE5,
     constraints: {
       voices: ELEVENLABS_VOICES,
       languages: tts.languages,
@@ -3785,7 +3786,7 @@ function geminiTts(id, label, description) {
     adapter: "fal",
     providerModelId,
     lifecycle: "active",
-    exposure: EXPOSURE4,
+    exposure: EXPOSURE5,
     constraints: {
       voices: GEMINI_TTS_VOICES,
       supports: ["style"],
@@ -3808,7 +3809,7 @@ function syncLipsync(id, label, providerModelId, syncModel, description) {
     adapter: "fal",
     providerModelId,
     lifecycle: "active",
-    exposure: EXPOSURE4,
+    exposure: EXPOSURE5,
     constraints: {
       requires: ["sourceVideo", "sourceAudio"],
       references: LIPSYNC_AUDIO,
@@ -3869,7 +3870,7 @@ var MEDIA_NODE_MODELS = [
     adapter: "fal",
     providerModelId: "fal-ai/minimax/speech-2.8-hd",
     lifecycle: "active",
-    exposure: EXPOSURE4,
+    exposure: EXPOSURE5,
     constraints: {
       voices: MINIMAX_VOICES,
       languages: MINIMAX_SPEECH_LANGUAGES,
@@ -3894,7 +3895,7 @@ var MEDIA_NODE_MODELS = [
     adapter: "fal",
     providerModelId: "fal-ai/elevenlabs/voice-changer",
     lifecycle: "active",
-    exposure: EXPOSURE4,
+    exposure: EXPOSURE5,
     constraints: {
       requires: ["sourceAudio"],
       // The flat price assumes at most a minute of speech.
@@ -3921,7 +3922,7 @@ var MEDIA_NODE_MODELS = [
     adapter: "fal",
     providerModelId: "fal-ai/elevenlabs/dubbing",
     lifecycle: "active",
-    exposure: EXPOSURE4,
+    exposure: EXPOSURE5,
     constraints: {
       requiresOneOf: ["sourceVideo", "sourceAudio"],
       // The flat price covers 60 s of source.
@@ -3969,7 +3970,7 @@ var MEDIA_NODE_MODELS = [
     adapter: "fal",
     providerModelId: "veed/lipsync/v2",
     lifecycle: "active",
-    exposure: EXPOSURE4,
+    exposure: EXPOSURE5,
     constraints: {
       requires: ["sourceVideo", "sourceAudio"],
       references: LIPSYNC_AUDIO,
@@ -3995,7 +3996,7 @@ var MEDIA_NODE_MODELS = [
     adapter: "fal",
     providerModelId: "minimax/h3-max/lip-sync/image-to-video",
     lifecycle: "active",
-    exposure: EXPOSURE4,
+    exposure: EXPOSURE5,
     constraints: {
       requires: ["firstFrame", "sourceAudio"],
       resolutions: ["480p", "768p", "1080p", "2k"],
@@ -4021,7 +4022,7 @@ var MEDIA_NODE_MODELS = [
     adapter: "fal",
     providerModelId: "fal-ai/elevenlabs/speech-to-text/scribe-v2",
     lifecycle: "active",
-    exposure: EXPOSURE4,
+    exposure: EXPOSURE5,
     constraints: {
       requires: ["sourceAudio"],
       // The flat price covers 10 minutes of audio.
@@ -5636,7 +5637,7 @@ var VIDEO_FAMILY_MODELS = [
 
 // src/catalog/video-luma.ts
 var VIDEO_USAGE2 = { unit: "seconds" };
-var EXPOSURE5 = ["cowork", "space_builder"];
+var EXPOSURE6 = ["cowork", "space_builder"];
 var LUMA_LOGO2 = "https://cdn.jsdelivr.net/npm/@lobehub/icons-static-png@1.97.0/light/luma-color.png";
 var RAY_RATIOS = ["9:16", "3:4", "1:1", "4:3", "16:9", "21:9"];
 var RAY_RESOLUTIONS = ["360p", "720p", "1080p"];
@@ -5654,7 +5655,7 @@ var LUMA_VIDEO_MODELS = [
     adapter: "luma",
     providerModelId: "ray-3.2",
     lifecycle: "active",
-    exposure: EXPOSURE5,
+    exposure: EXPOSURE6,
     constraints: {
       aspectRatios: ["adaptive", ...RAY_RATIOS],
       resolutions: [...RAY_RESOLUTIONS],
@@ -5686,7 +5687,7 @@ var LUMA_VIDEO_MODELS = [
     adapter: "luma",
     providerModelId: "ray-3.2",
     lifecycle: "active",
-    exposure: EXPOSURE5,
+    exposure: EXPOSURE6,
     constraints: {
       resolutions: [...RAY_RESOLUTIONS],
       requires: ["sourceVideo"],
@@ -5719,7 +5720,7 @@ var LUMA_VIDEO_MODELS = [
     adapter: "luma",
     providerModelId: "ray-3.2",
     lifecycle: "active",
-    exposure: EXPOSURE5,
+    exposure: EXPOSURE6,
     constraints: {
       aspectRatios: [...RAY_RATIOS],
       resolutions: [...RAY_RESOLUTIONS],
@@ -6409,6 +6410,48 @@ var VIDEO_MODELS = [
     },
     capabilities: { ...FRAMES_EXCLUDE_REFERENCES, lastFrameRequiresFirstFrame: true }
   },
+  {
+    // fal's Wan 3.0 Prime endpoints share Wan 3.0's input schemas field for field but bill
+    // more per second, so Prime is its own row (the id is the pricing key).
+    // https://fal.ai/models/alibaba/wan-3.0-prime/reference-to-video/api
+    id: "wan-3.0-prime",
+    aliases: ["wan-3-prime", "wan3-prime"],
+    label: "Wan 3 Prime",
+    type: "video",
+    adapter: "fal",
+    providerModelId: "alibaba/wan-3.0-prime",
+    lifecycle: "active",
+    exposure: ["cowork", "space_builder"],
+    constraints: {
+      aspectRatios: [...WAN_RATIOS],
+      resolutions: ["480p", "720p", "1080p"],
+      durations: { min: 2, max: 30 },
+      durationPresets: [5, 10, 15, 30],
+      references: {
+        images: 10,
+        videos: 5,
+        audios: 5,
+        videoSeconds: { total: 15 },
+        audioSeconds: { total: 15 }
+      },
+      supports: ["seed", "lastFrame", "generateAudio", "safetyChecker"],
+      prompt: WAN_PROMPT,
+      audio: "optional"
+    },
+    defaults: {
+      durationSeconds: 5,
+      resolution: "1080p",
+      aspectRatio: "adaptive",
+      generateAudio: true
+    },
+    usage: { ...VIDEO_USAGE3 },
+    pricingKind: "video_per_second",
+    display: {
+      provider: "alibaba",
+      description: "Alibaba Wan 3 Prime, the faster-generating tier of Wan 3, for text-to-video, image-to-video (start/end frame), and reference-to-video from up to 10 images, 5 videos and 5 audio clips, with native audio; 2-30 seconds at up to 1080p."
+    },
+    capabilities: { ...FRAMES_EXCLUDE_REFERENCES, lastFrameRequiresFirstFrame: true }
+  },
   {
     // Duration 3-10s per trigger-workflows mapDuration; the endpoint takes no resolution.
     id: "kling-o1-v2v",