@picsart/ai-sdk 6.15.1 → 6.17.0

package/index.d.ts

@@ -349,14 +349,30 @@ type ModelInputById = {
     "gemini-2.5-flash-tts": {
         language?: string;
         accent?: string;
-        prompt: string;
+        prompt?: string;
         voiceId?: string;
+        parts?: Array<{
+            text: string;
+            speaker?: string;
+        }>;
+        multiSpeakerVoiceConfigs?: Array<{
+            speaker: string;
+            voiceName: string;
+        }>;
     };
     "gemini-2.5-pro-tts": {
         language?: string;
         accent?: string;
-        prompt: string;
+        prompt?: string;
         voiceId?: string;
+        parts?: Array<{
+            text: string;
+            speaker?: string;
+        }>;
+        multiSpeakerVoiceConfigs?: Array<{
+            speaker: string;
+            voiceName: string;
+        }>;
     };
     "gemini-3-pro": {
         prompt: string;
@@ -412,14 +428,34 @@ type ModelInputById = {
     "gemini-3.8-flash-lite-tts": {
         language?: string;
         accent?: string;
-        prompt: string;
+        prompt?: string;
         voiceId?: string;
+        style?: string;
+        parts?: Array<{
+            text: string;
+            speaker?: string;
+            style?: string;
+        }>;
+        multiSpeakerVoiceConfigs?: Array<{
+            speaker: string;
+            voiceName: string;
+        }>;
     };
     "gemini-3.8-flash-tts": {
         language?: string;
         accent?: string;
-        prompt: string;
+        prompt?: string;
         voiceId?: string;
+        style?: string;
+        parts?: Array<{
+            text: string;
+            speaker?: string;
+            style?: string;
+        }>;
+        multiSpeakerVoiceConfigs?: Array<{
+            speaker: string;
+            voiceName: string;
+        }>;
     };
     "gemini-omni-1.1-flash-preview": {
         prompt: string;
@@ -689,6 +725,25 @@ type ModelInputById = {
         negativePrompt?: string;
         cfgScale?: number;
     };
+    "ideogram-4-5": {
+        prompt: string;
+        imageUrls?: string[];
+        mask?: string;
+        size?: "auto" | "source" | "2048x2048" | "1440x2880" | "2880x1440" | "1664x2496" | "2496x1664" | "1792x2240" | "2240x1792" | "1440x2560" | "2560x1440" | "1600x2560" | "2560x1600" | "1728x2304" | "2304x1728" | "1296x3168" | "3168x1296" | "1152x2944" | "2944x1152" | "1248x3328" | "3328x1248" | "1280x3072" | "3072x1280" | "1024x3072" | "3072x1024" | "1024x1024" | "896x1120" | "1120x896" | "864x1152" | "1152x864" | "832x1248" | "1248x832" | "800x1280" | "1280x800" | "720x1280" | "1280x720" | "720x1440" | "1440x720" | "512x1536" | "1536x512";
+        magicPrompt?: "auto" | "on" | "off";
+        quality?: "low" | "medium" | "high";
+        count?: 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8;
+        seed?: number;
+    };
+    "ideogram-4-5-precise-edit": {
+        prompt: string;
+        startFrame: string;
+        mask?: string;
+        imageUrls?: string[];
+        quality?: "low" | "medium" | "high" | "very_high";
+        count?: 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8;
+        seed?: number;
+    };
     "ideogram-character": {
         prompt: string;
         resolution?: "1024x1024" | "1344x768" | "768x1344" | "1152x864" | "864x1152" | "832x1248" | "1280x800";
@@ -3249,6 +3304,8 @@ declare const Models: {
     readonly HeygenTalkingPhoto: "heygen-talking-photo";
     readonly HeygenVideoAvatar: "heygen-video-avatar";
     readonly HunyuanV3: "hunyuan-v3";
+    readonly Ideogram45: "ideogram-4-5";
+    readonly Ideogram45PreciseEdit: "ideogram-4-5-precise-edit";
     readonly IdeogramCharacter: "ideogram-character";
     readonly IdeogramPImage: "ideogram-p-image";
     readonly IdeogramV3: "ideogram-v3";

package/index.js

@@ -6453,11 +6453,6 @@ var buildGemini31FlashLiteImagePayload = (ctx) => ({
   imageSize: ctx.resolution ?? "1K",
   ...buildThinkingConfig(ctx)
 });
-var buildGeminiTTSPayload = (model) => (ctx) => ({
-  text: ctx.prompt,
-  model,
-  voiceName: ctx.voiceId ?? "Kore"
-});
 function inferMimeType2(url) {
   return url.match(/\.png(\?|$)/i) ? "image/png" : "image/jpeg";
 }
@@ -6469,6 +6464,52 @@ var buildGeminiOmniVideoPayload = (ctx) => ({
   ...ctx.imageUrls?.[0] ? { image: { url: ctx.imageUrls[0], mimeType: inferMimeType2(ctx.imageUrls[0]) } } : {},
   ...ctx.videoUrl ? { video: { url: ctx.videoUrl } } : {}
 });
+var GEMINI_TTS_TEXT_MAX = 6e3;
+var GEMINI_TTS_STYLE_MAX = 500;
+var geminiTtsParams = (directable) => ({
+  ...params.language(true),
+  // 6,000 boundary-verified against the live gemini/v1/audios worker
+  // (accepts >5,000; see scripts/api-tests/audio-charlimit-boundary-probe.mjs).
+  ...params.prompt({ required: false, maxLength: GEMINI_TTS_TEXT_MAX }),
+  ...params.voiceId([], GEMINI_DEFAULT_VOICE_ID, { catalog: { workflow: "gemini/v1/catalog/voices" } }),
+  ...directable ? {
+    style: {
+      label: "Style",
+      descriptor: {
+        kind: "text",
+        maxLength: GEMINI_TTS_STYLE_MAX,
+        placeholder: "e.g. warm, unhurried; a bedtime story"
+      }
+    }
+  } : {},
+  // Worker `SpeechPartDto[]`, same names. Used instead of `prompt`.
+  parts: {
+    label: "Spoken Parts",
+    descriptor: {
+      kind: "object",
+      array: { min: 1, max: 200 },
+      fields: {
+        text: { label: "Text", kind: "text", maxLength: GEMINI_TTS_TEXT_MAX },
+        speaker: { label: "Speaker", kind: "text", required: false },
+        ...directable ? { style: { label: "Style", kind: "text", maxLength: GEMINI_TTS_STYLE_MAX, required: false } } : {}
+      }
+    }
+  },
+  // Worker `SpeakerVoiceConfigDto[]`, max 2. Replaces `voiceId`. The voice is
+  // a plain id, not a catalog picker: catalog loading resolves top-level params
+  // only, so a caller lists voices with `ai.catalogs.voices(<model>)`.
+  multiSpeakerVoiceConfigs: {
+    label: "Speakers",
+    descriptor: {
+      kind: "object",
+      array: { min: 1, max: 2 },
+      fields: {
+        speaker: { label: "Speaker", kind: "text", placeholder: "Speaker 1" },
+        voiceName: { label: "Voice", kind: "text", placeholder: GEMINI_DEFAULT_VOICE_ID }
+      }
+    }
+  }
+});
 var GEMINI_AR_WIDE = ["1:1", "16:9", "9:16", "3:4", "4:3", "2:3", "21:9", "auto"];
 var thinkingLevelParam = {
   thinkingLevel: {
@@ -6586,32 +6627,35 @@ var { MODELS: MODELS22 } = defineModels("google", [
     }
   },
   // ── Audio ─────────────────────────────────────────────────────────
+  // Payload builders live in gemini.payloads.ts (typed against ModelInput).
+  // The wire `model` is fixed per entry there: `modelId` is the pricing key and
+  // never reaches the builder.
+  //
+  // 2.5 and 3.8 take direction and speakers differently, so the builders
+  // resolve them per generation. 3.8 reads `style` and per-part speakers from
+  // `speechMetadata` and speaks a natural-language prefix in the text aloud:
+  // `parts` go through and `language` / `accent` fold into `style`. 2.5
+  // ignores `speechMetadata`: it reads speakers from "Speaker: line" labels
+  // and direction from a "Say …:" prefix, so the builder writes both into
+  // one text.
   {
     id: "gemini-2.5-flash-tts",
     name: "Gemini 2.5 Flash TTS",
     addedAt: "2026-02-15",
     workflow: "gemini/v1/audios",
-    buildPayload: buildGeminiTTSPayload("gemini-2.5-flash-tts"),
     estimatedTime: 15,
     mode: "audio",
     inputType: "tts",
     modelId: "gemini-2.5-flash-tts",
     description: "Google Gemini native text-to-speech with expressive multilingual voices.",
-    features: [feat("Multilingual", "characteristic"), feat("30 Voices", "characteristic")],
-    paramConfig: {
-      ...params.language(true),
-      // 6,000 boundary-verified against the live gemini/v1/audios worker
-      // (accepts >5,000; see scripts/api-tests/audio-charlimit-boundary-probe.mjs).
-      ...params.prompt({ maxLength: 6e3 }),
-      ...params.voiceId([], GEMINI_DEFAULT_VOICE_ID, { catalog: { workflow: "gemini/v1/catalog/voices" } })
-    }
+    features: [feat("Multilingual", "characteristic"), feat("30 Voices", "characteristic"), feat("Multi-Speaker", "characteristic")],
+    paramConfig: geminiTtsParams(false)
   },
   {
     id: "gemini-2.5-pro-tts",
     name: "Gemini 2.5 Pro TTS",
     addedAt: "2026-03-18",
     workflow: "gemini/v1/audios",
-    buildPayload: buildGeminiTTSPayload("gemini-2.5-pro-tts"),
     estimatedTime: 20,
     mode: "audio",
     inputType: "tts",
@@ -6619,54 +6663,34 @@ var { MODELS: MODELS22 } = defineModels("google", [
     badge: ["premium"],
     description: "Premium Gemini TTS with richer expressiveness and multi-speaker support.",
     features: [feat("Multilingual", "characteristic"), feat("30 Voices", "characteristic"), feat("Multi-Speaker", "characteristic")],
-    paramConfig: {
-      ...params.language(true),
-      // 6,000 boundary-verified against the live gemini/v1/audios worker
-      // (accepts >5,000; see scripts/api-tests/audio-charlimit-boundary-probe.mjs).
-      ...params.prompt({ maxLength: 6e3 }),
-      ...params.voiceId([], GEMINI_DEFAULT_VOICE_ID, { catalog: { workflow: "gemini/v1/catalog/voices" } })
-    }
+    paramConfig: geminiTtsParams(false)
   },
   {
     id: "gemini-3.8-flash-tts",
     name: "Gemini 3.8 Flash TTS",
     addedAt: "2026-09-24",
     workflow: "gemini/v1/audios",
-    buildPayload: buildGeminiTTSPayload("gemini-3.8-flash-tts"),
     estimatedTime: 15,
     mode: "audio",
     inputType: "tts",
     modelId: "gemini-3.8-flash-tts",
-    description: "Latest Gemini native text-to-speech with expressive multilingual voices.",
-    features: [feat("Multilingual", "characteristic"), feat("30 Voices", "characteristic")],
-    paramConfig: {
-      ...params.language(true),
-      // 6,000 boundary-verified against the live gemini/v1/audios worker
-      // (accepts >5,000; see scripts/api-tests/audio-charlimit-boundary-probe.mjs).
-      ...params.prompt({ maxLength: 6e3 }),
-      ...params.voiceId([], GEMINI_DEFAULT_VOICE_ID, { catalog: { workflow: "gemini/v1/catalog/voices" } })
-    }
+    description: "Latest Gemini native text-to-speech with directable delivery and inline vocal events like <laugh> and <sigh>.",
+    features: [feat("Multilingual", "characteristic"), feat("30 Voices", "characteristic"), feat("Multi-Speaker", "characteristic"), feat("Style Direction", "characteristic")],
+    paramConfig: geminiTtsParams(true)
   },
   {
     id: "gemini-3.8-flash-lite-tts",
     name: "Gemini 3.8 Flash Lite TTS",
     addedAt: "2026-09-24",
     workflow: "gemini/v1/audios",
-    buildPayload: buildGeminiTTSPayload("gemini-3.8-flash-lite-tts"),
     estimatedTime: 12,
     mode: "audio",
     inputType: "tts",
     modelId: "gemini-3.8-flash-lite-tts",
     badge: ["fast"],
-    description: "Lightweight Gemini 3.8 TTS variant for faster, high-volume speech synthesis.",
-    features: [feat("Multilingual", "characteristic"), feat("30 Voices", "characteristic")],
-    paramConfig: {
-      ...params.language(true),
-      // 6,000 boundary-verified against the live gemini/v1/audios worker
-      // (accepts >5,000; see scripts/api-tests/audio-charlimit-boundary-probe.mjs).
-      ...params.prompt({ maxLength: 6e3 }),
-      ...params.voiceId([], GEMINI_DEFAULT_VOICE_ID, { catalog: { workflow: "gemini/v1/catalog/voices" } })
-    }
+    description: "Lightweight Gemini 3.8 TTS with directable delivery for faster, high-volume speech synthesis.",
+    features: [feat("Multilingual", "characteristic"), feat("30 Voices", "characteristic"), feat("Multi-Speaker", "characteristic"), feat("Style Direction", "characteristic")],
+    paramConfig: geminiTtsParams(true)
   },
   // ── Video ─────────────────────────────────────────────────────────
   {
@@ -6733,8 +6757,95 @@ var buildOmniFlash11Payload = (input) => ({
   ...input.videoUrl ? { video: { url: input.videoUrl } } : {},
   ...input.videoUrls?.length ? { referenceVideos: input.videoUrls.map((url) => ({ url })) } : {}
 });
+var STYLE_MAX = 500;
+var TEXT_MAX = 6e3;
+var invalid = (message) => new ApiError(message, { status: 400, code: "validation_error" });
+var blank = (value) => !value?.trim();
+function checkInput(input) {
+  const configs = input.multiSpeakerVoiceConfigs ?? [];
+  for (const [i, config] of configs.entries()) {
+    if (blank(config.speaker)) throw invalid(`Gemini TTS: speaker config ${i + 1} has no speaker.`);
+    if (blank(config.voiceName)) throw invalid(`Gemini TTS: speaker config ${i + 1} has no voiceName.`);
+  }
+  const speakers = configs.map((config) => config.speaker);
+  if (new Set(speakers).size !== speakers.length) {
+    throw invalid("Gemini TTS: multiSpeakerVoiceConfigs names the same speaker twice.");
+  }
+  const parts = input.parts?.length ? input.parts : void 0;
+  if (!parts) {
+    if (blank(input.prompt)) throw invalid("Gemini TTS: provide a prompt or parts.");
+    return { speakers };
+  }
+  for (const [i, part] of parts.entries()) {
+    if (blank(part.text)) throw invalid(`Gemini TTS: part ${i + 1} has no text.`);
+    if (speakers.length && (!part.speaker || !speakers.includes(part.speaker))) {
+      throw invalid(
+        `Gemini TTS: part ${i + 1} needs a speaker from multiSpeakerVoiceConfigs (${speakers.join(", ")}).`
+      );
+    }
+    if (!speakers.length && part.speaker) {
+      throw invalid(`Gemini TTS: part ${i + 1} names a speaker but no multiSpeakerVoiceConfigs are set.`);
+    }
+  }
+  return { parts, speakers };
+}
+function assertLength(spoken) {
+  if (spoken.length > TEXT_MAX) {
+    throw invalid(`Gemini TTS: the text totals ${spoken.length} characters; the limit is ${TEXT_MAX}.`);
+  }
+}
+var voiceFields = (input) => input.multiSpeakerVoiceConfigs?.length ? { multiSpeakerVoiceConfigs: input.multiSpeakerVoiceConfigs } : { voiceName: input.voiceId ?? GEMINI_DEFAULT_VOICE_ID };
+function composeStyle(input) {
+  const language = input.language?.trim();
+  const accent = input.accent?.trim();
+  const voice = language || accent ? `speak${language ? ` in ${language}` : ""}${accent ? ` with a ${accent} accent` : ""}` : void 0;
+  const composed = [...input.styles.map((style) => style?.trim()), voice].filter(Boolean).join("; ") || void 0;
+  if (composed && composed.length > STYLE_MAX) {
+    throw invalid(`Gemini TTS: style with language/accent exceeds ${STYLE_MAX} characters.`);
+  }
+  return composed;
+}
+function directionPrefix(input) {
+  const language = input.language?.trim();
+  const accent = input.accent?.trim();
+  if (!language && !accent) return "";
+  return `Say${language ? ` in ${language}` : ""}${accent ? ` with a ${accent} accent` : ""}: `;
+}
+var buildTts25Payload = (model) => (input) => {
+  const { parts, speakers } = checkInput(input);
+  const body = parts ? parts.map((part) => speakers.length ? `${part.speaker}: ${part.text}` : part.text).join("\n") : input.prompt;
+  const text = `${directionPrefix(input)}${body}`;
+  assertLength(text);
+  return { text, model, ...voiceFields(input) };
+};
+var buildTts38Payload = (model) => (input) => {
+  const { parts } = checkInput(input);
+  const direction = { language: input.language, accent: input.accent };
+  if (!parts) {
+    assertLength(input.prompt);
+    const style = composeStyle({ styles: [input.style], ...direction });
+    return { text: input.prompt, model, ...voiceFields(input), ...style ? { style } : {} };
+  }
+  assertLength(parts.map((part) => part.text).join("\n"));
+  return {
+    parts: parts.map((part) => {
+      const style = composeStyle({ styles: [input.style, part.style], ...direction });
+      return {
+        text: part.text,
+        ...part.speaker ? { speaker: part.speaker } : {},
+        ...style ? { style } : {}
+      };
+    }),
+    model,
+    ...voiceFields(input)
+  };
+};
 registerPayloads(MODELS22, {
-  "gemini-omni-1.1-flash-preview": buildOmniFlash11Payload
+  "gemini-omni-1.1-flash-preview": buildOmniFlash11Payload,
+  "gemini-2.5-flash-tts": buildTts25Payload("gemini-2.5-flash-tts"),
+  "gemini-2.5-pro-tts": buildTts25Payload("gemini-2.5-pro-tts"),
+  "gemini-3.8-flash-tts": buildTts38Payload("gemini-3.8-flash-tts"),
+  "gemini-3.8-flash-lite-tts": buildTts38Payload("gemini-3.8-flash-lite-tts")
 });
 
 // src/vendors/catalog/openai.ts
@@ -7691,7 +7802,138 @@ var buildIdeogramPImagePayload = (ctx) => ({
   resolution: ctx.resolution ?? "1024x1024",
   rendering_speed: ctx.renderingSpeed ?? "medium"
 });
+var IDEOGRAM_45_IMAGE_BOUNDS = {
+  maxBytes: 25 * 1024 * 1024,
+  minAspectRatio: 1 / 6,
+  maxAspectRatio: 6
+};
+var IDEOGRAM_45_T2I_SIZES = [
+  "2048x2048",
+  "1440x2880",
+  "2880x1440",
+  "1664x2496",
+  "2496x1664",
+  "1792x2240",
+  "2240x1792",
+  "1440x2560",
+  "2560x1440",
+  "1600x2560",
+  "2560x1600",
+  "1728x2304",
+  "2304x1728",
+  "1296x3168",
+  "3168x1296",
+  "1152x2944",
+  "2944x1152",
+  "1248x3328",
+  "3328x1248",
+  "1280x3072",
+  "3072x1280",
+  "1024x3072",
+  "3072x1024",
+  "1024x1024",
+  "896x1120",
+  "1120x896",
+  "864x1152",
+  "1152x864",
+  "832x1248",
+  "1248x832",
+  "800x1280",
+  "1280x800",
+  "720x1280",
+  "1280x720",
+  "720x1440",
+  "1440x720",
+  "512x1536",
+  "1536x512"
+];
+var ideogram45MaskParam = p.file("mask", "image", {
+  label: "Mask",
+  category: "asset",
+  ...IDEOGRAM_45_IMAGE_BOUNDS
+});
+var ideogram45MagicPromptParam = {
+  magicPrompt: {
+    label: "Magic Prompt",
+    descriptor: {
+      kind: "enum",
+      valueType: "string",
+      options: [
+        { id: "auto", label: "Auto" },
+        { id: "on", label: "On" },
+        { id: "off", label: "Off (verbatim)" }
+      ],
+      default: "auto"
+    }
+  }
+};
+var IDEOGRAM_45_COUNTS = [1, 2, 3, 4, 5, 6, 7, 8];
+var ideogram45Constraints = [
+  { when: { imageUrls: { exists: false } }, then: {
+    size: { allowed: ["auto", ...IDEOGRAM_45_T2I_SIZES] },
+    mask: { disabled: true, reason: "A mask needs a source image." }
+  } },
+  { when: { imageUrls: { exists: true } }, then: {
+    size: { allowed: ["auto", "source"], reason: 'With input images the size is "auto" or the source image size.' }
+  } },
+  { when: { mask: { exists: true } }, then: {
+    size: { disabled: true, reason: "A masked edit keeps the source image size." }
+  } }
+];
+var ideogram45PreciseEditConstraints = [
+  { when: { quality: { is: "very_high" } }, then: {
+    count: { allowed: [1, 2, 3, 4], reason: "Very high quality returns at most 4 images." }
+  } }
+];
 var { MODELS: MODELS26 } = defineModels("ideogram", [
+  {
+    id: "ideogram-4-5",
+    name: "Ideogram 4.5",
+    addedAt: "2026-09-28",
+    release: "preview",
+    workflow: "ideogram/v4.5/generate",
+    estimatedTime: 30,
+    mode: "image",
+    inputType: "t2i",
+    description: "Ideogram 4.5 \u2014 text-to-image and drift-free iterative editing with up to 4 reference images and an optional mask.",
+    features: [feat("Reference Images", "input"), feat("Masked Edit", "input"), feat("2K", "resolution")],
+    constraints: ideogram45Constraints,
+    paramConfig: {
+      ...params.prompt({ maxLength: 1e4 }),
+      // imageUrls[0] is the source image to edit; the rest are references ("image 2", "image 3"…).
+      // With a mask the vendor allows at most 4 images.
+      ...params.imageInput(5, "Images", false, "reference", IDEOGRAM_45_IMAGE_BOUNDS),
+      ...ideogram45MaskParam,
+      ...p.size(["auto", "source", ...IDEOGRAM_45_T2I_SIZES], "auto"),
+      ...ideogram45MagicPromptParam,
+      ...p.quality(["low", "medium", "high"], "high"),
+      ...params.count(IDEOGRAM_45_COUNTS, 1),
+      ...params.seed()
+    }
+  },
+  {
+    id: "ideogram-4-5-precise-edit",
+    name: "Ideogram 4.5 Precise Edit",
+    addedAt: "2026-09-28",
+    release: "preview",
+    workflow: "ideogram/v4.5/precise-edit",
+    estimatedTime: 40,
+    mode: "image",
+    inputType: "i2i",
+    description: "Ideogram 4.5 precise edit \u2014 change one area and keep the rest of the image pixel-intact, optionally with a mask and reference images.",
+    features: [feat("Precise Edit", "input"), feat("Masked Edit", "input"), feat("Reference Images", "input")],
+    constraints: ideogram45PreciseEditConstraints,
+    paramConfig: {
+      ...params.prompt({ maxLength: 1e4 }),
+      ...params.startFrame("Source Image", true, IDEOGRAM_45_IMAGE_BOUNDS),
+      ...ideogram45MaskParam,
+      // With a mask the vendor allows at most 3 references.
+      ...params.imageInput(4, "Reference Images", false, "reference", IDEOGRAM_45_IMAGE_BOUNDS),
+      ...p.quality(["low", "medium", "high", "very_high"], "high"),
+      ...params.count(IDEOGRAM_45_COUNTS, 1),
+      ...params.seed()
+    }
+  },
   {
     id: "ideogram-v4",
     name: "Ideogram 4.0",
@@ -7888,6 +8130,69 @@ var { MODELS: MODELS26 } = defineModels("ideogram", [
   }
 ]);
 
+// src/vendors/catalog/ideogram.payloads.ts
+var GENERATE_MAX_IMAGES_WITH_MASK = 4;
+var PRECISE_EDIT_MAX_REFERENCES_WITH_MASK = 3;
+var VERY_HIGH_MAX_COUNT = 4;
+function assertMaskedImageLimit(model, count, max, what) {
+  if (count > max) {
+    throw new ApiError(`${model}: with a mask, use at most ${max} ${what} (got ${count}).`, {
+      status: 400,
+      code: "validation_error"
+    });
+  }
+}
+function ideogram45Size(size, hasImages, hasMask) {
+  if (!size || hasMask) return void 0;
+  if (hasImages) return size === "auto" || size === "source" ? size : void 0;
+  return size === "source" ? void 0 : size;
+}
+var buildIdeogram45Payload = (input) => {
+  const images = input.imageUrls?.length ? input.imageUrls : void 0;
+  const mask = images && input.mask ? input.mask : void 0;
+  if (mask) assertMaskedImageLimit("Ideogram 4.5", images.length, GENERATE_MAX_IMAGES_WITH_MASK, "images");
+  const size = ideogram45Size(input.size, !!images, !!mask);
+  return {
+    prompt: input.prompt,
+    ...images ? { images } : {},
+    ...mask ? { mask } : {},
+    ...size ? { size } : {},
+    ...input.magicPrompt ? { magic_prompt: input.magicPrompt } : {},
+    quality: input.quality ?? "high",
+    num_images: input.count ?? 1,
+    ...input.seed != null ? { seed: input.seed } : {}
+  };
+};
+var buildIdeogram45PreciseEditPayload = (input) => {
+  if (input.mask) {
+    assertMaskedImageLimit(
+      "Ideogram 4.5 Precise Edit",
+      input.imageUrls?.length ?? 0,
+      PRECISE_EDIT_MAX_REFERENCES_WITH_MASK,
+      "reference images"
+    );
+  }
+  if (input.quality === "very_high" && (input.count ?? 1) > VERY_HIGH_MAX_COUNT) {
+    throw new ApiError(
+      `Ideogram 4.5 Precise Edit: very_high quality returns at most ${VERY_HIGH_MAX_COUNT} images (got ${input.count}).`,
+      { status: 400, code: "validation_error" }
+    );
+  }
+  return {
+    image: input.startFrame,
+    prompt: input.prompt,
+    ...input.mask ? { mask: input.mask } : {},
+    ...input.imageUrls?.length ? { reference_images: input.imageUrls } : {},
+    quality: input.quality ?? "high",
+    num_images: input.count ?? 1,
+    ...input.seed != null ? { seed: input.seed } : {}
+  };
+};
+registerPayloads(MODELS26, {
+  "ideogram-4-5": buildIdeogram45Payload,
+  "ideogram-4-5-precise-edit": buildIdeogram45PreciseEditPayload
+});
+
 // src/vendors/catalog/qwen.ts
 var QWEN_V1_SIZES = [
   "2048x2048",
@@ -13002,6 +13307,8 @@ var Happyhorse11T2v = "happyhorse-1.1-t2v";
 var HeygenTalkingPhoto = "heygen-talking-photo";
 var HeygenVideoAvatar = "heygen-video-avatar";
 var HunyuanV3 = "hunyuan-v3";
+var Ideogram45 = "ideogram-4-5";
+var Ideogram45PreciseEdit = "ideogram-4-5-precise-edit";
 var IdeogramCharacter = "ideogram-character";
 var IdeogramPImage = "ideogram-p-image";
 var IdeogramV3 = "ideogram-v3";
@@ -13255,6 +13562,8 @@ var Models = {
   HeygenTalkingPhoto,
   HeygenVideoAvatar,
   HunyuanV3,
+  Ideogram45,
+  Ideogram45PreciseEdit,
   IdeogramCharacter,
   IdeogramPImage,
   IdeogramV3,

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@picsart/ai-sdk",
-  "version": "6.15.1",
+  "version": "6.17.0",
   "type": "module",
   "description": "Type-safe SDK for 100+ AI models — image, video, audio, and text generation with Picsart",
   "license": "MIT",