@howells/motif-sdk 1.2.0 → 1.3.0

package/dist/image.js

@@ -19,10 +19,10 @@ var MODELS = {
   gpt2: {
     benchmark: {
       artificialAnalysis: {
-        editing: { elo: 1256, pricePer1k: 211, rank: 3 },
+        editing: { elo: 1256, rank: 3 },
         snapshotDate: AA_IMAGE_LEADERBOARD_SNAPSHOT,
         sourceUrls: AA_IMAGE_SOURCES,
-        textToImage: { elo: 1369, pricePer1k: 211, rank: 1 }
+        textToImage: { elo: 1369, rank: 1 }
       },
       speed: {
         medianSeconds: 200.7,
@@ -62,10 +62,10 @@ var MODELS = {
   gpt: {
     benchmark: {
       artificialAnalysis: {
-        editing: { elo: 1250, pricePer1k: 133, rank: 5 },
+        editing: { elo: 1250, rank: 5 },
         snapshotDate: AA_IMAGE_LEADERBOARD_SNAPSHOT,
         sourceUrls: AA_IMAGE_SOURCES,
-        textToImage: { elo: 1310, pricePer1k: 133, rank: 4 }
+        textToImage: { elo: 1310, rank: 4 }
       },
       speed: {
         medianSeconds: 34.6,
@@ -105,10 +105,10 @@ var MODELS = {
   banana2: {
     benchmark: {
       artificialAnalysis: {
-        editing: { elo: 1249, pricePer1k: 67, rank: 7 },
+        editing: { elo: 1249, rank: 7 },
         snapshotDate: AA_IMAGE_LEADERBOARD_SNAPSHOT,
         sourceUrls: AA_IMAGE_SOURCES,
-        textToImage: { elo: 1320, pricePer1k: 67, rank: 3 }
+        textToImage: { elo: 1320, rank: 3 }
       },
       tiers: { price: "premium", quality: "frontier", speed: "unknown" },
       useCase: "Best balance of quality, edit support, web search, and cost"
@@ -146,10 +146,10 @@ var MODELS = {
   banana: {
     benchmark: {
       artificialAnalysis: {
-        editing: { elo: 1245, pricePer1k: 134, rank: 9 },
+        editing: { elo: 1245, rank: 9 },
         snapshotDate: AA_IMAGE_LEADERBOARD_SNAPSHOT,
         sourceUrls: AA_IMAGE_SOURCES,
-        textToImage: { elo: 1297, pricePer1k: 134, rank: 6 }
+        textToImage: { elo: 1297, rank: 6 }
       },
       speed: {
         medianSeconds: 19.2,
@@ -191,7 +191,7 @@ var MODELS = {
   gemini: {
     benchmark: {
       artificialAnalysis: {
-        editing: { elo: 1180, pricePer1k: 39, rank: 29 },
+        editing: { elo: 1180, rank: 29 },
         snapshotDate: AA_IMAGE_LEADERBOARD_SNAPSHOT,
         sourceUrls: AA_IMAGE_SOURCES
       }
@@ -225,10 +225,10 @@ var MODELS = {
   gemini3: {
     benchmark: {
       artificialAnalysis: {
-        editing: { elo: 1245, pricePer1k: 134, rank: 9 },
+        editing: { elo: 1245, rank: 9 },
         snapshotDate: AA_IMAGE_LEADERBOARD_SNAPSHOT,
         sourceUrls: AA_IMAGE_SOURCES,
-        textToImage: { elo: 1297, pricePer1k: 134, rank: 6 }
+        textToImage: { elo: 1297, rank: 6 }
       }
     },
     editEndpoint: "fal-ai/gemini-3-pro-image-preview/edit",
@@ -261,10 +261,10 @@ var MODELS = {
   seedream4: {
     benchmark: {
       artificialAnalysis: {
-        editing: { elo: 1185, pricePer1k: 30, rank: 28 },
+        editing: { elo: 1185, rank: 28 },
         snapshotDate: AA_IMAGE_LEADERBOARD_SNAPSHOT,
         sourceUrls: AA_IMAGE_SOURCES,
-        textToImage: { elo: 1226, pricePer1k: 30, rank: 16 }
+        textToImage: { elo: 1226, rank: 16 }
       },
       speed: {
         medianSeconds: 14.3,
@@ -301,10 +301,10 @@ var MODELS = {
   seedream45: {
     benchmark: {
       artificialAnalysis: {
-        editing: { elo: 1186, pricePer1k: 40, rank: 26 },
+        editing: { elo: 1186, rank: 26 },
         snapshotDate: AA_IMAGE_LEADERBOARD_SNAPSHOT,
         sourceUrls: AA_IMAGE_SOURCES,
-        textToImage: { elo: 1203, pricePer1k: 40, rank: 32 }
+        textToImage: { elo: 1203, rank: 32 }
       },
       speed: {
         medianSeconds: 18.2,
@@ -341,10 +341,10 @@ var MODELS = {
   seedream5: {
     benchmark: {
       artificialAnalysis: {
-        editing: { elo: 1248, pricePer1k: 90, rank: 8 },
+        editing: { elo: 1248, rank: 8 },
         snapshotDate: AA_IMAGE_LEADERBOARD_SNAPSHOT,
         sourceUrls: AA_IMAGE_SOURCES,
-        textToImage: { elo: 1279, pricePer1k: 90, rank: 10 }
+        textToImage: { elo: 1279, rank: 10 }
       }
     },
     name: "Seedream 5.0 Pro",
@@ -375,7 +375,7 @@ var MODELS = {
   "seedream5-lite": {
     benchmark: {
       artificialAnalysis: {
-        editing: { elo: 1169, pricePer1k: 35, rank: 32 },
+        editing: { elo: 1169, rank: 32 },
         snapshotDate: AA_IMAGE_LEADERBOARD_SNAPSHOT,
         sourceUrls: AA_IMAGE_SOURCES
       }
@@ -408,10 +408,10 @@ var MODELS = {
   "flux2-max": {
     benchmark: {
       artificialAnalysis: {
-        editing: { elo: 1201, pricePer1k: 140, rank: 19 },
+        editing: { elo: 1201, rank: 19 },
         snapshotDate: AA_IMAGE_LEADERBOARD_SNAPSHOT,
         sourceUrls: AA_IMAGE_SOURCES,
-        textToImage: { elo: 1225, pricePer1k: 70, rank: 17 }
+        textToImage: { elo: 1225, rank: 17 }
       },
       speed: {
         medianSeconds: 31.6,
@@ -451,10 +451,10 @@ var MODELS = {
   "flux2-pro": {
     benchmark: {
       artificialAnalysis: {
-        editing: { elo: 1170, pricePer1k: 45, rank: 31 },
+        editing: { elo: 1170, rank: 31 },
         snapshotDate: AA_IMAGE_LEADERBOARD_SNAPSHOT,
         sourceUrls: AA_IMAGE_SOURCES,
-        textToImage: { elo: 1207, pricePer1k: 30, rank: 30 }
+        textToImage: { elo: 1207, rank: 30 }
       },
       speed: {
         medianSeconds: 16.1,
@@ -496,7 +496,7 @@ var MODELS = {
       artificialAnalysis: {
         snapshotDate: AA_IMAGE_LEADERBOARD_SNAPSHOT,
         sourceUrls: AA_IMAGE_SOURCES,
-        textToImage: { elo: 1223, pricePer1k: 60, rank: 18 }
+        textToImage: { elo: 1223, rank: 18 }
       },
       speed: {
         medianSeconds: 15.3,
@@ -689,10 +689,10 @@ var MODELS = {
         // #2 on BOTH boards, which almost nothing else manages - the top text-to-image model
         // (GPT Image 2) drops to 3rd for editing, and the top editor (MAI-Image-2.5-Pro) is not
         // on fal at all. Reve is the best editor motif can actually reach.
-        editing: { elo: 1263, pricePer1k: 200, rank: 2 },
+        editing: { elo: 1263, rank: 2 },
         snapshotDate: AA_IMAGE_LEADERBOARD_SNAPSHOT,
         sourceUrls: AA_IMAGE_SOURCES,
-        textToImage: { elo: 1322, pricePer1k: 200, rank: 2 }
+        textToImage: { elo: 1322, rank: 2 }
       },
       tiers: { price: "premium", quality: "frontier", speed: "unknown" },
       useCase: "Highest-ranked editor reachable on fal; strong prompt adherence and text rendering"
@@ -729,7 +729,7 @@ var MODELS = {
       artificialAnalysis: {
         snapshotDate: AA_IMAGE_LEADERBOARD_SNAPSHOT,
         sourceUrls: AA_IMAGE_SOURCES,
-        textToImage: { elo: 1215, pricePer1k: 35, rank: 25 }
+        textToImage: { elo: 1215, rank: 25 }
       }
     },
     endpoint: "fal-ai/recraft/v4/text-to-image",
@@ -786,7 +786,7 @@ var MODELS = {
       artificialAnalysis: {
         snapshotDate: AA_IMAGE_LEADERBOARD_SNAPSHOT,
         sourceUrls: AA_IMAGE_SOURCES,
-        textToImage: { elo: 1220, pricePer1k: 60, rank: 19 }
+        textToImage: { elo: 1220, rank: 19 }
       }
     },
     name: "Ideogram V4",
@@ -821,10 +821,10 @@ var MODELS = {
   "grok-image": {
     benchmark: {
       artificialAnalysis: {
-        editing: { elo: 1214, pricePer1k: 20, rank: 16 },
+        editing: { elo: 1214, rank: 16 },
         snapshotDate: AA_IMAGE_LEADERBOARD_SNAPSHOT,
         sourceUrls: AA_IMAGE_SOURCES,
-        textToImage: { elo: 1216, pricePer1k: 20, rank: 23 }
+        textToImage: { elo: 1216, rank: 23 }
       },
       speed: {
         medianSeconds: 5.1,
@@ -898,10 +898,10 @@ var MODELS = {
   qwen3: {
     benchmark: {
       artificialAnalysis: {
-        editing: { elo: 1218, pricePer1k: 33, rank: 15 },
+        editing: { elo: 1218, rank: 15 },
         snapshotDate: AA_IMAGE_LEADERBOARD_SNAPSHOT,
         sourceUrls: AA_IMAGE_SOURCES,
-        textToImage: { elo: 1272, pricePer1k: 30, rank: 11 }
+        textToImage: { elo: 1272, rank: 11 }
       },
       tiers: { price: "budget", quality: "better", speed: "unknown" },
       useCase: "Strong quality per dollar - 11th on text-to-image at $30/1k"
@@ -1335,6 +1335,13 @@ var ASSET_TOOLS = {
     inputKind: "image",
     name: "Seedream v5 Pro Layerize",
     outputKeys: ["layers", "images"],
+    // Every element of `layers` carries its own `name` ("Left amber glass
+    // bottle") and a `z_index` that orders the stack, verified live. Without
+    // this the stack lands as layers.png … layers-5.png and the only way to
+    // tell the reconstructed background plate from an object is to open them.
+    outputLabels: {
+      layers: { fromItem: { nameField: "name", orderField: "z_index" } }
+    },
     price: { kind: "metered" },
     pricing: "$0.03375 per generated layer below 1536x1536 total area, $0.0675 per layer above; the listed rate is per layer, not per call",
     queued: true,