{ "$schema": "https://json-schema.org/draft/2020-12/schema", "_comment": "Ground truth for LetzAI model routing, pricing, and provider endpoints.", "_pricingSchema": { "_description": "Pricing calculation rules", "image": "Use mode key (1k, 2k, hd, 4k, uhd) or _flat for fixed price", "video": "Use _flat OR (_perSecond × duration). Apply multipliers: _audioMultiplier (if audio enabled), _hqMultiplier (if HQ enabled). Multiplier defaults to 1 if not specified. Special: _withAudio overrides flat+audio cost.", "upscaling": "Use size key (e.g. '1k','2k','4k') or scale-factor key (e.g. '2x','4x','8x') or _flat for fixed price." }, "models": { "gemini-3-pro-image-preview": { "active": true, "displayName": "Nano Banana Pro", "description": "Google's state-of-the-art image generation model with improved text rendering, multi-turn editing, and professional-grade controls over lighting, camera, and composition. Served via Vertex AI.", "dateAdded": "2025-11-20", "provider": "vertex", "allowBYOK": true, "aliases": [ "gemini", "google/gemini", "gemini-3-pro-image-preview" ], "capabilities": { "generation": { "providerModel": "gemini-3-pro-image-preview", "pricing": { "1k": 80, "2k": 160, "hd": 160, "4k": 240, "uhd": 240 }, "defaultMode": "1k", "nsfwCheck": false, "notes": "Built-in safety filter; no external NSFW needed" }, "editing": { "providerModel": "gemini-3-pro-image-preview", "pricing": { "1k": 80, "2k": 160, "hd": 160, "4k": 240, "uhd": 240 }, "defaultMode": "1k", "nsfwCheck": false } }, "fallback": "nano-banana-pro", "isDefault": true, "_byokNote": "Vertex backend; nbp-inferencesh uses Vertex for frontend" }, "gemini-2.5-flash-image": { "active": false, "displayName": "Gemini 2.5 Flash Image", "description": "Google's fast and cost-effective image generation model with support for 10 aspect ratios and image-only output capabilities.", "dateAdded": "2025-08-26", "provider": "vertex", "allowBYOK": true, "aliases": ["gemini-2.5-flash-image"], "capabilities": { "generation": { "providerModel": "gemini-2.5-flash-image", "pricing": { "_flat": 50 }, "nsfwCheck": false, "notes": "Flat pricing regardless of mode" }, "editing": { "providerModel": "gemini-2.5-flash-image", "pricing": { "_flat": 50 }, "nsfwCheck": false } }, "frontend": { "value": "nano-banana", "label": "Nano Banana", "description": "Great Image Editing Model from Google", "image": "/blog/en-US/image-models/cinematic.png", "providerIcon": "/providers/google.webp", "category": "Third-Party", "baseModel": "gemini-2.5-flash-image", "defaultMode": "default", "resolutions": ["1k"], "resolutionModes": { "1k": "default" }, "pricingModes": { "1k": "1k" }, "showInGenerate": false, "showInEdit": true, "order": 3 }, "fallback": "nano-banana-pro", "_byokNote": "Routed through InferenceSH when USEVERTEX=FALSE" }, "seedream": { "active": true, "displayName": "Seedream 4.5", "description": "ByteDance's image model with multi-image consistency, excellent text rendering, and up to 4K resolution. Maintains character identity across multiple images.", "dateAdded": "2025-12-04", "provider": "byteplus", "allowBYOK": false, "aliases": [ "seedream", "seedream-4.5", "seedream-4-5", "seedream-4-5-251128", "bytedance-seedream-4.5" ], "capabilities": { "generation": { "providerModel": "ep-20251208175106-nbz5k", "pricing": { "1k": 80, "2k": 80, "hd": 80, "4k": 160, "uhd": 160 }, "defaultMode": "2k", "nsfwCheck": true, "notes": "External NSFW check via InferenceSH" }, "editing": { "providerModel": "ep-20251208175106-nbz5k", "pricing": { "1k": 80, "2k": 80, "hd": 80, "4k": 160, "uhd": 160 }, "defaultMode": "2k", "nsfwCheck": true } }, "frontend": { "value": "seedream-4-5", "label": "Seedream 4.5", "description": "Powerful Image Generation Model from Bytedance", "image": "/blog/en-US/image-models/cinematic.png", "providerIcon": "/providers/bytedance.png", "category": "Third-Party", "baseModel": "seedream-4-5-251128", "defaultMode": "2k", "resolutions": ["hd", "4k"], "resolutionModes": { "hd": "2k", "4k": "4k" }, "showInGenerate": true, "showInEdit": true, "order": 4 } }, "flux-2": { "active": true, "displayName": "Flux 2", "description": "Black Forest Labs' 32B parameter model with multi-reference image support, 4MP editing, improved text rendering, and enhanced photorealism.", "dateAdded": "2025-11-25", "provider": "fal", "allowBYOK": false, "aliases": [ "flux2", "flux-2", "flux-2-max", "flux-2-pro", "flux-2-dev" ], "capabilities": { "generation": { "providerModel": "fal-ai/flux-2-max", "pricing": { "1k": 60, "hd": 120 }, "defaultMode": "1k", "nsfwCheck": false, "notes": "FAL built-in has_nsfw_concepts" }, "editing": { "providerModel": "fal-ai/flux-2-max/edit", "pricing": { "1k": 60, "hd": 120 }, "defaultMode": "1k", "nsfwCheck": false } }, "frontend": { "value": "flux2-max", "label": "Flux2 [max]", "description": "Highest quality generation", "image": "/blog/en-US/image-models/default.png", "providerIcon": "/providers/bfl.jpg", "category": "Third-Party", "baseModel": "flux2", "defaultMode": "1k", "resolutions": ["1k", "hd"], "resolutionModes": { "1k": "1k", "hd": "hd" }, "showInGenerate": true, "showInEdit": true, "order": 5 } }, "gpt-image-2": { "active": true, "displayName": "GPT Image 2", "description": "OpenAI's GPT Image 2 via FAL — exceptional typography, fine-detail rendering, and precise mask-based editing up to 4K.", "dateAdded": "2026-04-21", "provider": "fal", "allowBYOK": false, "aliases": [ "gpt-image-2", "gpt-image2", "gptimage2", "openai/gpt-image-2" ], "capabilities": { "generation": { "providerModel": "openai/gpt-image-2", "pricing": { "1k": 160, "2k": 240, "hd": 240, "4k": 480, "uhd": 480 }, "defaultMode": "1k", "nsfwCheck": false, "notes": "FAL openai/gpt-image-2; quality locked to 'high'" }, "editing": { "providerModel": "openai/gpt-image-2/edit", "pricing": { "1k": 160, "2k": 240, "hd": 240, "4k": 480, "uhd": 480 }, "defaultMode": "1k", "nsfwCheck": false, "notes": "FAL openai/gpt-image-2/edit; supports image_urls + optional mask_url" } }, "frontend": { "value": "gpt-image-2", "label": "GPT Image 2", "description": "OpenAI's latest image model with exceptional typography", "image": "/blog/en-US/image-models/default.png", "providerIcon": "/providers/chatgpt.png", "providerName": "OpenAI", "category": "Third-Party", "baseModel": "gpt-image-2", "defaultMode": "1k", "resolutions": ["1k", "hd", "4k"], "resolutionModes": { "1k": "1k", "hd": "hd", "4k": "4k" }, "showInGenerate": true, "showInEdit": true, "order": 3 } }, "flux-kontext-lora": { "active": false, "displayName": "Flux Kontext LoRA", "description": "Black Forest Labs' in-context image editing model with LoRA support, enabling iterative multi-turn editing with character and style preservation.", "dateAdded": "2025-05-29", "provider": "fal", "allowBYOK": false, "aliases": [], "capabilities": { "editing": { "providerModel": "fal-ai/flux-kontext-lora", "pricing": { "1k": 80, "hd": 160 }, "defaultMode": "1k", "nsfwCheck": false, "notes": "Auto-selected when LoRAs are present; not directly requestable by alias" } } }, "_comment_video_edit": "=== VIDEO EDITING MODELS ===", "veo-3.1-extend": { "active": true, "displayName": "Veo 3.1 Extend Video", "description": "Google's video extension model capable of extending existing videos up to 148 seconds with maintained consistency and style.", "dateAdded": "2025-10-15", "provider": "vertex", "allowBYOK": true, "aliases": [ "veo31extend", "veo3.1extend", "veo-3.1-extend", "veo31-extend" ], "capabilities": { "video_editing": { "providerModel": "veo-3.1-generate-001", "pricing": { "_perSecond": 10 }, "duration": { "min": 1, "max": 8 }, "nsfwCheck": false, "notes": "Video extension via Vertex AI; credits = duration_seconds * 10; clamped 1-8s" } }, "frontend": { "value": "veo-3.1", "label": "Google VEO 3.1", "providerIcon": "/providers/google.webp", "providerName": "Google", "description": "Video extension model by Google", "supportedModes": ["extend"], "order": 2 }, "jobType": "video_edit" }, "kling-o3": { "active": true, "displayName": "Kling O3 Pro Edit", "description": "Kuaishou's unified multimodal video editor with 7-in-1 capabilities including object removal, background swapping, style changes, and character consistency.", "dateAdded": "2026-02-05", "provider": "fal", "allowBYOK": false, "aliases": ["klingo3", "kling-o3", "kling_o3", "kling-o3-edit"], "capabilities": { "video_editing": { "providerModel": "fal-ai/kling-video/o3/pro/video-to-video/edit", "pricing": { "_flat": 50 }, "nsfwCheck": false, "notes": "Video-to-video edit; flat pricing" } }, "frontend": { "value": "kling-o3", "label": "Kling O3", "providerIcon": "/providers/kling.png", "providerName": "Kuaishou", "description": "Video editing model by Kling", "supportedModes": ["edit"], "order": 1 }, "jobType": "video_edit" }, "beeble-switchx": { "active": true, "displayName": "Beeble SwitchX", "description": "Beeble's video-to-video model that swaps the background or scene in a clip while preserving the original subject, motion, framing and lighting. Not for character or outfit changes. Driven by an optional reference image plus prompt.", "dateAdded": "2026-05-18", "provider": "beeble", "allowBYOK": false, "aliases": ["beeble", "switchx", "beeble-switchx", "switch-x"], "capabilities": { "video_editing": { "providerModel": "switchx", "pricing": { "_flat": 50 }, "nsfwCheck": false, "maxFrames": 240, "maxSourcePixels": 2770000, "notes": "Video-to-video background/scene swap only (no character or outfit changes); flat credits same as Kling O3 Pro Edit. Beeble caps: ≤240 frames, ≤2.77M source pixels." } }, "frontend": { "value": "beeble-switchx", "label": "Beeble SwitchX", "providerIcon": "/providers/beeble.png", "providerName": "Beeble", "description": "Background / scene swap by Beeble", "supportedModes": ["edit"], "order": 3 }, "jobType": "video_edit" }, "nbp-inferencesh": { "active": true, "displayName": "Nano Banana Pro", "description": "Google Gemini 3 Pro via InferenceSH with improved retries and per-organization API key support.", "dateAdded": "2025-11-20", "provider": "vertex", "allowBYOK": true, "aliases": ["nbp-inferencesh", "nbp-inf", "nano-banana-pro-inf"], "capabilities": { "generation": { "providerModel": "gemini-3-pro-image-preview", "pricing": { "1k": 80, "2k": 160, "hd": 160, "4k": 240, "uhd": 240 }, "defaultMode": "1k", "nsfwCheck": false, "notes": "Gemini 3 Pro via Vertex AI; supports per-org API key override" }, "editing": { "providerModel": "gemini-3-pro-image-preview", "pricing": { "1k": 80, "2k": 160, "hd": 160, "4k": 240, "uhd": 240 }, "defaultMode": "1k", "nsfwCheck": false } }, "frontend": { "value": "nano-banana-pro", "label": "Nano Banana Pro", "description": "The latest Image Generation Model from Google", "image": "/blog/en-US/image-models/cinematic.png", "providerIcon": "/providers/google.webp", "category": "Third-Party", "baseModel": "gemini-3-pro-image-preview", "defaultMode": "default", "resolutions": ["1k", "hd", "4k"], "resolutionModes": { "1k": "default", "hd": "2k", "4k": "4k" }, "pricingModes": { "1k": "1k", "hd": "2k", "4k": "4k" }, "showInGenerate": true, "showInEdit": true, "order": 1 }, "fallback": "nano-banana-pro", "isDefault": true, "_byokNote": "Always uses InferenceSH; org key used when orgId is present" }, "gemini-3-1-flash-image-preview": { "active": true, "displayName": "Nano Banana 2", "description": "Google's Gemini 3.1 Flash image generation model — fast, cost-effective generation and editing with up to 4K resolution, advanced text rendering, and Google Search grounding.", "dateAdded": "2026-02-26", "provider": "vertex", "allowBYOK": true, "aliases": [ "gemini-3-1-flash", "gemini-3.1-flash-image-preview", "gemini-3-1-flash-image-preview" ], "capabilities": { "generation": { "providerModel": "gemini-3.1-flash-image-preview", "pricing": { "1k": 40, "2k": 80, "hd": 80, "4k": 160, "uhd": 160 }, "defaultMode": "1k", "nsfwCheck": false, "notes": "Gemini 3.1 Flash via Vertex AI; built-in safety filter" }, "editing": { "providerModel": "gemini-3.1-flash-image-preview", "pricing": { "1k": 50, "2k": 100, "hd": 100, "4k": 150, "uhd": 150 }, "defaultMode": "1k", "nsfwCheck": false } }, "fallback": "nano-banana-pro", "_byokNote": "Vertex backend; nbf-inferencesh uses InferenceSH for frontend" }, "nbf-inferencesh": { "active": true, "displayName": "Nano Banana 2", "description": "Google Gemini 3.1 Flash via InferenceSH — fast, cost-effective image generation and editing with up to 4K resolution and Google Search grounding.", "dateAdded": "2026-02-26", "provider": "inferencesh", "allowBYOK": true, "aliases": ["nbf-inferencesh", "nbf-inf", "nano-banana-flash-inf"], "capabilities": { "generation": { "providerModel": "gemini-3.1-flash-image-preview", "providerApp": "google/gemini-3-1-flash-image-preview@0c7ma1ex", "pricing": { "1k": 40, "2k": 80, "hd": 80, "4k": 160, "uhd": 160 }, "defaultMode": "1k", "nsfwCheck": false, "notes": "Gemini 3.1 Flash via InferenceSH; supports per-org API key override" }, "editing": { "providerModel": "gemini-3.1-flash-image-preview", "providerApp": "google/gemini-3-1-flash-image-preview@0c7ma1ex", "pricing": { "1k": 50, "2k": 100, "hd": 100, "4k": 150, "uhd": 150 }, "defaultMode": "1k", "nsfwCheck": false } }, "frontend": { "value": "nano-banana-2", "label": "Nano Banana 2", "description": "Fast Image Generation Model from Google", "image": "/blog/en-US/image-models/cinematic.png", "providerIcon": "/providers/google.webp", "category": "Third-Party", "baseModel": "nbf-inferencesh", "defaultMode": "default", "resolutions": ["1k", "hd", "4k"], "resolutionModes": { "1k": "default", "hd": "2k", "4k": "4k" }, "pricingModes": { "1k": "1k", "hd": "2k", "4k": "4k" }, "showInGenerate": true, "showInEdit": true, "order": 2 }, "_byokNote": "Uses InferenceSH; org key used when orgId is present" }, "phota": { "active": false, "displayName": "Phota", "description": "Image generation and editing via InferenceSH. When @handle references are present, routes through the edit endpoint for identity-aware composition.", "dateAdded": "2026-03-26", "provider": "inferencesh", "allowBYOK": false, "aliases": ["phota", "photo-edit"], "capabilities": { "generation": { "providerModel": "phota", "providerApp": "phota/generate@7geytb17", "pricing": { "1k": 80, "4k": 240 }, "defaultMode": "1k", "nsfwCheck": false, "notes": "Text-to-image generation; when @handle references are present the edit app is used instead. Supports 1-4 output images." }, "editing": { "providerModel": "phota", "providerApp": "phota/edit@2kpwhdga", "pricing": { "1k": 80, "4k": 240 }, "defaultMode": "1k", "nsfwCheck": false, "notes": "Image editing with 1-10 input images; supports 1-4 output images" } }, "frontend": { "value": "phota", "label": "Phota", "description": "Image generation and editing with subject consistency", "image": "/blog/en-US/image-models/cinematic.png", "providerIcon": "/providers/inferencesh.webp", "providerName": "Phota", "category": "Third-Party", "baseModel": "phota", "defaultMode": "default", "resolutions": ["1k", "4k"], "resolutionModes": { "1k": "default", "4k": "4k" }, "pricingModes": { "1k": "1k", "4k": "4k" }, "showInGenerate": true, "showInEdit": true, "order": 4 } }, "wan-2-7-image-pro": { "active": true, "displayName": "WAN 2.7 Image Pro", "description": "Alibaba's professional image generation model supporting text-to-image, image editing, and multi-reference generation with up to 4K high-definition output. Supports thinking mode for improved quality.", "dateAdded": "2026-04-01", "provider": "inferencesh", "allowBYOK": false, "aliases": [ "wan27", "wan-2-7", "wan-2.7", "wan-2-7-image-pro", "wan-2.7-image-pro", "wan27-image-pro" ], "capabilities": { "generation": { "providerModel": "wan-2-7-image-pro", "providerApp": "alibaba/wan-2-7-image-pro@7394ycc2", "pricing": { "1k": 10, "2k": 10, "hd": 10, "4k": 10, "uhd": 10 }, "defaultMode": "2k", "nsfwCheck": false, "notes": "WAN 2.7 Image Pro via InferenceSH; supports up to 9 reference images, thinking mode, sequential image sets; up to 4K" }, "editing": { "providerModel": "wan-2-7-image-pro", "providerApp": "alibaba/wan-2-7-image-pro@7394ycc2", "pricing": { "1k": 10, "2k": 10, "hd": 10 }, "defaultMode": "1k", "nsfwCheck": false, "notes": "Same app as generation; reference_images used for editing; max 2K for image-to-image" } }, "frontend": { "value": "wan-2-7", "label": "WAN 2.7 Pro", "description": "Professional Image Generation Model from Alibaba", "image": "/blog/en-US/image-models/cinematic.png", "providerIcon": "/providers/wan.png", "category": "Third-Party", "baseModel": "wan-2-7-image-pro", "defaultMode": "2k", "resolutions": ["1k", "hd"], "resolutionModes": { "1k": "1k", "hd": "2k" }, "pricingModes": { "1k": "1k", "hd": "2k" }, "showInGenerate": true, "showInEdit": true, "order": 5 } }, "grok-imagine": { "active": false, "displayName": "Grok Imagine Pro", "description": "xAI's image generation model with photorealistic rendering, excellent text and logo accuracy, and multi-style versatility from portraits to anime.", "dateAdded": "2026-03-12", "provider": "xai", "allowBYOK": false, "aliases": [ "grok-imagine", "grok-imagine-image", "grok-imagine-image-pro", "grok", "aurora", "xai" ], "capabilities": { "generation": { "providerModel": "grok-imagine-image-pro", "pricing": { "_flat": 50 }, "defaultMode": "1k", "nsfwCheck": true, "notes": "xAI Grok Imagine Pro via REST API; $0.07/image; supports aspect_ratio and resolution" }, "editing": { "providerModel": "grok-imagine-image-pro", "pricing": { "_flat": 50 }, "defaultMode": "1k", "nsfwCheck": true, "notes": "xAI Grok Imagine Pro edit via /images/edits; up to 3 input images; JSON body" } }, "frontend": { "value": "grok-imagine", "label": "Grok Imagine Pro", "description": "AI Image Generation by xAI", "image": "/blog/en-US/image-models/cinematic.png", "providerIcon": "/providers/xai.png", "category": "Third-Party", "baseModel": "grok-imagine-image-pro", "defaultMode": "1k", "resolutions": ["1k", "2k"], "resolutionModes": { "1k": "1k", "2k": "2k" }, "showInGenerate": true, "showInEdit": true, "order": 6 } }, "_comment_fallback": "=== FALLBACK MODELS ===", "nano-banana-pro": { "active": true, "displayName": "Nano Banana Pro (Fallback)", "description": "Fallback model via Fal.ai used when primary Gemini endpoints return 429 rate limit errors.", "dateAdded": "2025-11-20", "provider": "fal", "allowBYOK": false, "aliases": [], "capabilities": { "generation": { "providerModel": "fal-ai/nano-banana-pro", "editProviderModel": "fal-ai/nano-banana-pro/edit", "pricing": {}, "nsfwCheck": false, "notes": "Gemini 429 fallback only; no direct pricing (inherits from the model it fell back from)" } }, "isFallbackOnly": true }, "_comment_video": "=== VIDEO GENERATION MODELS (resolved by settings.mode) ===", "video-default": { "active": false, "displayName": "Default Video", "description": "Our first generation in-house default video model for video generation.", "dateAdded": "2025-07-01", "provider": "letzai", "allowBYOK": true, "aliases": ["default"], "capabilities": { "video": { "providerModel": "infsh/lightning-wan-2-2-i2v-a14b", "providerVersion": "06pqyf4ahanxy68c11b4fwcdba", "pricing": { "_perSecond": 60, "_audioMultiplier": 2, "_hqMultiplier": 2 }, "allowAudio": false, "supportsReference": false, "resolutions": ["720p", "1080p"], "duration": { "min": 2, "max": 6 }, "nsfwCheck": true, "notes": "Default video workflow; per-second pricing" } }, "jobType": "video_generation" }, "video-veo3": { "active": false, "displayName": "Google Veo 3.0", "description": "Google's video generation model via Vertex AI with native audio and cinematic quality at fixed 8-second duration.", "dateAdded": "2025-05-20", "provider": "vertex", "allowBYOK": false, "aliases": ["veo3"], "capabilities": { "video": { "providerModel": "veo-3.0-generate-001", "providerModelFast": "veo-3.0-fast-generate-001", "pricing": { "_flat": 1500, "_audioMultiplier": 2, "_hqMultiplier": 2 }, "allowAudio": true, "supportsReference": false, "duration": { "min": 8, "max": 8 }, "nsfwCheck": true, "resolutions": ["720p", "1080p"], "notes": "Google Veo 3.0 via Vertex AI; fixed 8s duration" } }, "jobType": "video_generation" }, "video-veo31": { "active": true, "displayName": "Google VEO3.1", "description": "Google's updated video model via Vertex AI with richer native audio, improved dialogue sync, and enhanced cinematic style.", "dateAdded": "2025-10-15", "provider": "vertex", "allowBYOK": false, "aliases": ["veo31"], "capabilities": { "video": { "providerModel": "veo-3.1-generate-001", "providerModelT2V": "veo-3.1-generate-001", "providerModelFast": "veo-3.1-fast-generate-001", "pricing": { "_flat": 1500, "_audioMultiplier": 2, "_hqMultiplier": 2 }, "allowAudio": true, "supportsReference": true, "duration": { "min": 8, "max": 8 }, "nsfwCheck": true, "resolutions": ["720p", "1080p"], "aspectRatios": ["original", "16:9", "9:16"], "features": { "multiFrame": true }, "notes": "Google Veo 3.1 via Vertex AI; fixed 8s duration; image-to-video (reference image); first/last frame; aspect ratios 16:9, 9:16, original" } }, "jobType": "video_generation" }, "video-veo31inf": { "active": false, "displayName": "Google VEO3.1", "description": "Google Veo 3.1 via Vertex AI with native audio, improved dialogue sync, and 8-second cinematic video generation.", "dateAdded": "2025-10-15", "provider": "vertex", "allowBYOK": true, "aliases": ["veo31inf"], "capabilities": { "video": { "providerModel": "veo-3.1-generate-001", "providerModelFast": "veo-3.1-fast-generate-001", "pricing": { "_flat": 1500, "_audioMultiplier": 2, "_hqMultiplier": 2 }, "allowAudio": true, "supportsReference": true, "duration": { "min": 8, "max": 8 }, "nsfwCheck": true, "resolutions": ["720p", "1080p"], "aspectRatios": ["original", "16:9", "9:16"], "notes": "Veo 3.1 via Vertex AI; aspect ratios 16:9, 9:16, original" } }, "jobType": "video_generation" }, "video-sora2": { "active": false, "displayName": "Sora 2", "description": "OpenAI's Sora 2 with custom characters, 16:9/9:16 exports, clips up to 20 seconds, video continuation, and image-to-video support.", "dateAdded": "2025-09-30", "provider": "openai", "allowBYOK": false, "aliases": ["sora2"], "capabilities": { "video": { "providerModel": "sora-2", "providerModelT2V": "sora-2", "providerModelPro": "sora-2-pro", "pricing": { "_flat": 100, "_audioMultiplier": 2, "_hqMultiplier": 2 }, "allowAudio": true, "supportsReference": true, "duration": { "min": 4, "max": 20 }, "nsfwCheck": true, "resolutions": ["720p", "1080p"], "aspectRatios": ["16:9", "9:16"], "notes": "OpenAI Sora 2 native API; input_reference for i2v; 4/8/12/16/20s; 1280x720, 720x1280, 1920x1080, 1080x1920" } }, "jobType": "video_generation" }, "video-kling25": { "active": false, "displayName": "Kling V2.5", "description": "Kuaishou's turbo video model optimized for faster generation with 5-10 second video support.", "dateAdded": "2025-08-01", "provider": "fal", "allowBYOK": false, "aliases": ["kling25"], "capabilities": { "video": { "providerModel": "fal-ai/kling-video/v2.5-turbo/pro/image-to-video", "pricing": { "_flat": 100, "_audioMultiplier": 2, "_hqMultiplier": 2 }, "allowAudio": false, "supportsReference": false, "duration": { "min": 5, "max": 10 }, "nsfwCheck": true, "resolutions": ["720p", "1080p"], "notes": "Kling V2.5 via Fal.ai" } }, "jobType": "video_generation" }, "video-kling3": { "active": true, "displayName": "Kling V3", "description": "Kuaishou's flagship video model with native multilingual audio, up to 15 seconds, multi-shot storytelling, and reference image support.", "dateAdded": "2026-02-05", "provider": "fal", "allowBYOK": false, "aliases": ["kling3", "kling30"], "capabilities": { "video": { "providerModel": "fal-ai/kling-video/v3/pro/image-to-video", "providerModelT2V": "fal-ai/kling-video/v3/pro/text-to-video", "imageArgKey": "start_image_url", "generateAudioArgKey": "generate_audio", "pricing": { "_flat": 150, "_withAudio": 300 }, "allowAudio": true, "supportsReference": true, "supportsElements": true, "duration": { "min": 3, "max": 15 }, "resolutions": ["1080p"], "aspectRatios": ["original", "16:9", "9:16"], "nsfwCheck": true, "features": { "multiPrompt": true, "multiFrame": true }, "multiPrompt": { "maxDuration": 15, "minShotDuration": 3 }, "notes": "Kling V3 via Fal.ai; supports reference images and elements; 1080p only; aspect ratios 16:9, 9:16, original; special audio pricing (300 credits with audio)" } }, "jobType": "video_generation" }, "video-kling26": { "active": true, "displayName": "Kling V2.6", "description": "Kuaishou's video model with simultaneous audio-visual generation, creating videos with voiceovers and sound effects in a single pass.", "dateAdded": "2025-12-03", "provider": "fal", "allowBYOK": false, "aliases": ["kling26"], "capabilities": { "video": { "providerModel": "fal-ai/kling-video/v2.6/pro/image-to-video", "providerModelT2V": "fal-ai/kling-video/v2.6/pro/text-to-video", "pricing": { "_flat": 150, "_audioMultiplier": 2, "_hqMultiplier": 2 }, "allowAudio": true, "supportsReference": false, "duration": { "min": 5, "max": 10, "step": 5 }, "nsfwCheck": true, "resolutions": ["720p", "1080p"], "aspectRatios": ["original", "16:9", "9:16"], "notes": "Kling V2.6 via Fal.ai; supports audio and text-to-video; aspect ratios 16:9, 9:16, original" } }, "jobType": "video_generation" }, "video-wan25": { "active": true, "displayName": "WAN 2.5", "description": "Alibaba's open-source video model with multilingual support, synchronized audio, and up to 10 seconds at 1080p resolution.", "dateAdded": "2025-09-24", "provider": "fal", "allowBYOK": false, "aliases": ["wan25"], "capabilities": { "video": { "providerModel": "fal-ai/wan-25-preview/image-to-video", "pricing": { "_flat": 110, "_audioMultiplier": 2, "_hqMultiplier": 2 }, "allowAudio": false, "supportsReference": false, "duration": { "min": 5, "max": 10 }, "nsfwCheck": true, "resolutions": ["720p", "1080p"], "aspectRatios": ["original", "16:9", "9:16"], "notes": "WAN 2.5 via Fal.ai; image-to-video only; aspect ratios 16:9, 9:16, original" } }, "jobType": "video_generation" }, "video-seedance2": { "active": true, "displayName": "Seedance 2.0", "description": "ByteDance's most advanced video model with cinematic output, native audio, real-world physics, and director-level camera control. Supports text, image, audio, and video reference inputs (up to 9 reference images, 3 reference videos, 3 reference audios) with first+last frame control.", "dateAdded": "2026-04-08", "provider": "inferencesh", "allowBYOK": false, "aliases": ["seedance2", "seedance20", "seedance-2"], "capabilities": { "video": { "providerModel": "bytedance/seedance-2-0", "providerModelT2V": "bytedance/seedance-2-0", "pricing": { "_perSecond": 300, "_perSecond480p": 180, "_perSecond720p": 300, "_perSecond1080p": 1200 }, "allowAudio": true, "alwaysAudio": true, "supportsReference": true, "supportsR2V": true, "duration": { "min": 4, "max": 15 }, "resolutions": ["480p", "720p", "1080p"], "aspectRatios": ["auto", "21:9", "16:9", "4:3", "1:1", "3:4", "9:16"], "features": { "multiFrame": true }, "notes": "Seedance 2.0 via InferenceSH (bytedance/seedance-2-0); single endpoint handles I2V/T2V/R2V via input fields (image/end_image for first+last frame; reference_images/reference_videos/reference_audios arrays for multimodal); per-second pricing (1200 credits/s @1080p, 300/s @720p, 180/s @480p); audio always included" } }, "frontend": { "value": "seedance2", "label": "Seedance 2.0", "providerIcon": "/providers/bytedance.png", "providerName": "ByteDance", "description": "Cinematic video with native audio", "supportedModes": ["generate"], "order": 3 }, "jobType": "video_generation" }, "video-seedance2-fast": { "active": false, "displayName": "Seedance 2.0 Fast", "description": "ByteDance's faster Seedance 2.0 Studio variant. Cinematic output with native audio at up to 720p, optimized for speed. Supports text, image, audio, and video reference inputs (up to 9 reference images, 3 reference videos, 3 reference audios) and first+last frame control.", "dateAdded": "2026-05-13", "provider": "inferencesh", "allowBYOK": false, "aliases": [ "seedance2-fast", "seedance20-fast", "seedance-2-fast", "seedance-2-0-fast", "seedance-2-0-studio-fast" ], "capabilities": { "video": { "providerModel": "bytedance/seedance-2-0-studio-fast", "providerModelT2V": "bytedance/seedance-2-0-studio-fast", "pricing": { "_perSecond": 150, "_perSecond480p": 90, "_perSecond720p": 150 }, "allowAudio": true, "alwaysAudio": true, "supportsReference": true, "supportsR2V": true, "duration": { "min": 4, "max": 15 }, "resolutions": ["480p", "720p"], "aspectRatios": ["auto", "21:9", "16:9", "4:3", "1:1", "3:4", "9:16"], "features": { "multiFrame": true }, "acceptsSafetyIdentifier": true, "notes": "Seedance 2.0 Fast via InferenceSH (bytedance/seedance-2-0-studio-fast); single endpoint handles I2V/T2V/R2V via input fields (image/end_image for first+last frame; reference_images/reference_videos/reference_audios arrays for multimodal); per-second pricing (150 credits/s @720p, 90/s @480p); audio always included; forwards safety_identifier when an org context is present (optional, not enforced)" } }, "frontend": { "value": "seedance2-fast", "label": "Seedance 2.0 Fast", "providerIcon": "/providers/bytedance.png", "providerName": "ByteDance", "description": "Fast cinematic video with native audio (up to 720p)", "supportedModes": ["generate"], "order": 4 }, "jobType": "video_generation" }, "video-seedance2-enterprise": { "active": true, "displayName": "Seedance 2.0 Enterprise", "description": "ByteDance's professional Seedance 2.0 Studio variant with private asset library support. Supports up to 1080p, with text-to-video, image-to-video, and multimodal reference-to-video (up to 9 reference images, 3 reference videos, 3 reference audios) and synchronized audio.", "dateAdded": "2026-05-13", "provider": "inferencesh", "allowBYOK": false, "aliases": [ "seedance2-enterprise", "seedance20-enterprise", "seedance-2-enterprise", "seedance-2-0-enterprise", "seedance-2-0-studio", "seedance2-studio" ], "enterpriseOnly": true, "capabilities": { "video": { "providerModel": "bytedance/seedance-2-0-studio", "providerModelT2V": "bytedance/seedance-2-0-studio", "pricing": { "_perSecond": 300, "_perSecond480p": 180, "_perSecond720p": 300, "_perSecond1080p": 1200 }, "allowAudio": true, "alwaysAudio": true, "supportsReference": true, "supportsR2V": true, "duration": { "min": 4, "max": 15 }, "resolutions": ["480p", "720p", "1080p"], "aspectRatios": ["auto", "21:9", "16:9", "4:3", "1:1", "3:4", "9:16"], "features": { "multiFrame": true }, "requireSafetyIdentifier": true, "acceptsSafetyIdentifier": true, "notes": "Seedance 2.0 Enterprise via InferenceSH (bytedance/seedance-2-0-studio); single endpoint handles I2V/T2V/R2V via input fields; per-second pricing (1200 credits/s @1080p, 300/s @720p, 180/s @480p); audio always included; enterprise only; supports up to 9 reference images, 3 reference videos, 3 reference audios with byteplus virtual portrait library. Requires a valid LetzAI organizationId (validated against /internal/organizations/{id} and forwarded as InferenceSH safety_identifier for per-org asset-library namespacing)." } }, "frontend": { "value": "seedance2-enterprise", "label": "Seedance 2.0 Enterprise", "providerIcon": "/providers/bytedance.png", "providerName": "ByteDance", "description": "Professional cinematic video with native audio (up to 1080p)", "supportedModes": ["generate"], "order": 5 }, "jobType": "video_generation" }, "video-grok": { "active": false, "displayName": "Grok Video", "description": "xAI's video generation model with text-to-video and image-to-video support, up to 15 seconds at 720p.", "dateAdded": "2026-03-12", "provider": "xai", "allowBYOK": false, "aliases": ["grok", "grok-video", "xai-video", "grok-imagine-video"], "capabilities": { "video": { "providerModel": "grok-imagine-video", "providerModelT2V": "grok-imagine-video", "pricing": { "_perSecond": 40, "_hqMultiplier": 2 }, "allowAudio": false, "supportsReference": true, "duration": { "min": 1, "max": 15 }, "nsfwCheck": true, "resolutions": ["480p", "720p"], "aspectRatios": ["16:9", "9:16", "1:1", "4:3", "3:4", "3:2", "2:3"], "notes": "xAI Grok Imagine Video via REST API; $0.050/second; async with polling; t2v and i2v" } }, "frontend": { "value": "grok-video", "label": "Grok Video", "providerIcon": "/providers/xai.png", "providerName": "xAI", "description": "Video generation by xAI", "supportedModes": ["generate"], "order": 5 }, "jobType": "video_generation" }, "_comment_upscale": "=== UPSCALING MODELS ===", "upscale-letzai": { "active": true, "displayName": "LetzAI Upscaler", "description": "LetzAI's in-house upscaler with three modes: Simple (high speed), Sharp (high quality), and Soft (high consistency).", "dateAdded": "2024-06-01", "provider": "comfyui", "allowBYOK": false, "aliases": ["simple", "sharp", "soft", "default"], "capabilities": { "upscaling": { "pricing": { "_flat": 20 }, "notes": "Local ComfyUI upscaler; modes: simple (SeedVR2), sharp (FLUX+UltraSharp), soft (FLUX+NMKD-Siax)" } }, "jobType": "upscale" }, "upscale-nano-banana-pro": { "active": true, "displayName": "Nano Banana Pro Upscaler", "description": "AI-powered upscaling using Google Gemini 3 Pro with intelligent detail enhancement up to 4K resolution.", "dateAdded": "2025-11-20", "provider": "inferencesh", "allowBYOK": false, "aliases": ["nano-banana-pro-upscale"], "capabilities": { "upscaling": { "pricing": { "1k": 80, "2k": 160, "4k": 240 }, "defaultMode": "2k", "notes": "Gemini 3 Pro via InferenceSH; mode value 'nano-banana-pro'; size 4→1K, 8→2K, 12→4K" } }, "frontend": { "value": "nano-banana-pro", "label": "Nano Banana Pro", "subLabel": "Up to 4K resolution", "providerIcon": "/providers/google.webp" }, "jobType": "upscale" }, "upscale-magnific-creative": { "active": true, "displayName": "Magnific Creative", "description": "Freepik's creative upscaler that can add and infer new details guided by text prompts, with up to 16x scaling.", "dateAdded": "2024-12-01", "provider": "freepik", "allowBYOK": false, "aliases": ["freepik", "magnific-creative"], "capabilities": { "upscaling": { "pricing": { "2x": 120, "4x": 240, "8x": 960, "16x": 960 }, "defaultMode": "4x", "notes": "Magnific Creative via Freepik API; creative upscale that can add/infer new details; supports prompt" } }, "frontend": { "value": "freepik", "label": "Magnific Creative", "subLabel": "Can add and infer new details", "providerIcon": "/providers/magnific.jpg" }, "jobType": "upscale" }, "upscale-magnific-precision": { "active": true, "displayName": "Magnific Precision V2", "description": "Freepik's faithful upscaler with no hallucinations — preserves text, UI elements, and original image structure with ultimate fidelity.", "dateAdded": "2025-11-10", "provider": "freepik", "allowBYOK": false, "aliases": ["freepik-precision", "magnific-precision"], "capabilities": { "upscaling": { "pricing": { "2x": 120, "4x": 240, "8x": 960, "16x": 960 }, "defaultMode": "4x", "notes": "Magnific Precision V2 via Freepik API; faithful upscale — no hallucinations, preserves text/UI; scale 2-16" } }, "frontend": { "value": "freepik-precision", "label": "Magnific Precision V2", "subLabel": "Faithful — preserves text & UI", "providerIcon": "/providers/magnific.jpg" }, "jobType": "upscale" }, "_comment_deprecated": "=== DEPRECATED MODELS ===", "letzai-v1": { "active": false, "displayName": "LetzAI V1", "description": "First generation LetzAI image model built on Stable Diffusion 1.5 with custom fine-tuning for creative generation.", "dateAdded": "2023-06-01", "provider": "comfyui", "allowBYOK": false, "aliases": ["letzai-v1", "letzaiv1"], "capabilities": { "generation": { "pricing": { "_flat": 5 }, "notes": "Deprecated - replaced by V2" } }, "deprecated": true }, "letzai-v2": { "active": false, "displayName": "LetzAI V2", "description": "Second generation LetzAI image model built on SDXL with improved detail, resolution, and prompt understanding.", "dateAdded": "2024-01-15", "provider": "comfyui", "allowBYOK": false, "aliases": ["letzai-v2", "letzaiv2"], "capabilities": { "generation": { "pricing": { "_flat": 10 }, "notes": "Deprecated - replaced by V3" } }, "deprecated": true }, "letzai-v3": { "active": false, "displayName": "LetzAI V3", "description": "Third generation LetzAI image model with enhanced quality and photorealism based on Flux1.dev.", "dateAdded": "2024-09-01", "provider": "comfyui", "allowBYOK": false, "aliases": ["letzai-v3", "letzaiv3"], "capabilities": { "generation": { "pricing": { "_flat": 15 }, "notes": "Deprecated - replaced by third-party models" } }, "deprecated": true } }, "providers": { "vertex": { "displayName": "Google Vertex AI", "envKeys": ["GOOGLE_CLOUD_PROJECT"], "notes": "Gemini models; can be toggled to InferenceSH via USEVERTEX=FALSE" }, "inferencesh": { "displayName": "InferenceSH", "envKeys": ["INF_API_TOKEN"], "supportsOrgKey": true, "notes": "Supports per-org API key override via orgId" }, "fal": { "displayName": "FAL.ai", "envKeys": ["FAL_KEY"], "notes": "Flux family, Kling, Nano Banana" }, "byteplus": { "displayName": "BytePlus", "envKeys": ["ARK_API_KEY"], "notes": "Seedream models" }, "xai": { "displayName": "xAI (Grok Imagine)", "envKeys": ["XAI_API_KEY"], "notes": "Grok Imagine image and video generation via Aurora model" }, "openai": { "displayName": "OpenAI", "envKeys": ["OPENAI_API_KEY"], "notes": "Sora 2 video generation via native Videos API" }, "freepik": { "displayName": "Freepik (Magnific)", "envKeys": ["FREEPIK_API_KEY"], "notes": "Magnific Creative and Precision V2 upscalers" }, "comfyui": { "displayName": "Local", "envKeys": [], "notes": "Local GPU workflows for LetzAI upscaling" } } }