{"routing":{"image":{"models":[{"id":16,"name":"Top-ranked","tier":"best","isDefault":true,"priceSats":1200,"description":"Top-ranked image generation (OpenAI GPT Image 2.5, ranked #1 on the Artificial Analysis text-to-image leaderboard as of September 2026, Elo 1197). 1200 sats flat, rendered at the maximum quality setting. Strict content filters."},{"id":8,"name":"Better","tier":"better","isDefault":false,"priceSats":150,"description":"High-detail image generation (ByteDance Seedream 5 Pro). 150 sats at 1K, 300 sats at 2K. Up to 10 reference images. The most permissive content policy of the three tiers."},{"id":2,"name":"Standard","tier":"standard","isDefault":false,"priceSats":100,"description":"Fast image generation (xAI Grok Imagine 2). 100 sats flat, 1K. Lowest cost option for simple prompts."}],"defaultModelId":16},"speech":{"models":[{"id":10,"name":"Transcription","tier":"best","isDefault":true,"priceSats":10,"description":"Transcribe audio to text. Supports 13 languages (en, zh, hi, es, ar, fr, pt, ru, de, ja, ko, it, nl). 10 sats/minute of audio."},{"id":33,"name":"Inworld TTS 1.5 Max","tier":"best","isDefault":true,"priceSats":1},{"id":3,"name":"Minimax Speech-2.8-HD","tier":null,"isDefault":false,"priceSats":1,"description":"Convert text to natural speech audio. Multiple voice options. 300 sats flat. Returns MP3."},{"id":43,"name":"OmniVoice TTS","tier":null,"isDefault":false,"priceSats":1}],"defaultModelId":33},"vision":{"models":[{"id":5,"name":"Vision Chat","tier":null,"isDefault":true,"priceSats":21,"description":"Analyze and describe image content. Ask questions about any image. 21 sats flat. Send image as base64."}],"defaultModelId":5},"text":{"models":[{"id":6,"name":"Kimi K3","tier":"best","isDefault":true,"priceSats":1,"description":"Best quality text model. 262K context window. Supports vision via imageBase64 param. Dynamic pricing: 100 chars/sat, minimum 15 sats."},{"id":31,"name":"Standard","tier":"standard","isDefault":false,"priceSats":1,"description":"Budget multilingual model. 119 languages accepted. Ultra-fast inference; top scorer in our translation evals. Dynamic pricing: 1000 chars/sat, minimum 5 sats. Best for translation and simple tasks."}],"defaultModelId":6},"Test":{"models":[{"id":9,"name":"Contact","tier":null,"isDefault":false,"priceSats":1}],"defaultModelId":null},"utility":{"models":[{"id":29,"name":"HTML to PDF","tier":"standard","isDefault":true,"priceSats":50},{"id":30,"name":"Receipt OCR","tier":"standard","isDefault":true,"priceSats":25},{"id":53,"name":"Render Card","tier":"standard","isDefault":false,"priceSats":5,"description":"Render a typographic card — headline, versus, list, stat or quote — as a PNG or JPEG. Deterministic layout, no AI model: the text you send is exactly what appears. For video slides, thumbnails and social cards."},{"id":11,"name":null,"tier":null,"isDefault":false,"priceSats":100},{"id":13,"name":null,"tier":null,"isDefault":false,"priceSats":100},{"id":26,"name":null,"tier":null,"isDefault":false,"priceSats":100},{"id":28,"name":null,"tier":null,"isDefault":false,"priceSats":200},{"id":46,"name":"E-Signature","tier":null,"isDefault":false,"priceSats":1000},{"id":47,"name":"Boarding Pass Wallet","tier":null,"isDefault":false,"priceSats":100}],"defaultModelId":30},"SearchAI":{"models":[{"id":15,"name":"Search","tier":null,"isDefault":false,"priceSats":21}],"defaultModelId":null},"3d":{"models":[{"id":17,"name":"Best","tier":"best","isDefault":true,"priceSats":1600,"description":"Convert a photo into a downloadable 3D GLB model. 1600 sats. Send image as base64 PNG/JPEG/WEBP. Async — may take up to 5 minutes."}],"defaultModelId":17},"image_edit":{"models":[{"id":18,"name":"Best","tier":null,"isDefault":false,"priceSats":200}],"defaultModelId":null},"telco":{"models":[{"id":20,"name":null,"tier":null,"isDefault":false,"priceSats":null,"pricing":"dynamic"}],"defaultModelId":null},"VoiceCloning":{"models":[{"id":22,"name":"Voice Cloning","tier":null,"isDefault":false,"priceSats":7500}],"defaultModelId":null},"music":{"models":[{"id":23,"name":"Music-3","tier":"best","isDefault":true,"priceSats":75,"description":"Generate a song from a style prompt plus lyrics (MiniMax Music-3). You choose the length: 15-300 seconds at 5 sats/second. Structure tags ([intro], [verse], [chorus], [bridge], [solo], [outro]) must each be on their own line; words on a tag line are dropped by the model. Tags with no words = instrumental. Returns 44.1 kHz stereo WAV. Prepaid - LNURL-withdraw refund on failure."}],"defaultModelId":23},"video_from_image":{"models":[{"id":24,"name":"Image to Video","tier":"best","isDefault":true,"priceSats":100,"description":"Image-to-video: animate a still image into cinematic video (MiniMax H3, top 3 on the Artificial Analysis image-to-video leaderboard). Provide an image + prompt. 5-15s, 768p or 2K (upscaled from a 768p render), native audio. Async - poll check_job_status."}],"defaultModelId":24},"video":{"models":[{"id":25,"name":"Video Generation","tier":"best","isDefault":true,"priceSats":50,"description":"Text-to-video generation (MiniMax H3, top 2 on the Artificial Analysis leaderboard). 5-15 second clips, 768p or 2K (2K is upscaled from a 768p render), native audio included free. Async - poll check_job_status after submitting."}],"defaultModelId":25},"ocr":{"models":[{"id":27,"name":"Mistral OCR","tier":"best","isDefault":true,"priceSats":10,"description":"Extract text from PDFs and images as clean Markdown. 10 sats per page. Send document as base64."}],"defaultModelId":27},"image_process":{"models":[{"id":35,"name":"BiRefNet","tier":"best","isDefault":true,"priceSats":44,"description":"Background removal using BiRefNet (state-of-the-art, Papers with Code). Handles hair, fur, glass, transparency, complex edges. Returns transparent PNG."},{"id":36,"name":"Real-ESRGAN","tier":"best","isDefault":true,"priceSats":5,"description":"Image upscaling 2x-4x using Real-ESRGAN (ICCV 2021, 100M+ production runs). Neural super-resolution that recovers real detail. Optional face enhancement."},{"id":37,"name":"CodeFormer","tier":"best","isDefault":true,"priceSats":25,"description":"Face restoration using CodeFormer (NeurIPS 2022, state-of-the-art on CelebA-Test). Restores blurry, damaged, or AI-generated faces. Adjustable fidelity parameter."},{"id":38,"name":"NSFW Detector","tier":"best","isDefault":true,"priceSats":2,"description":"NSFW image classification using fine-tuned ViT (87M+ runs). Returns 'normal' or 'nsfw'. 100x cheaper and faster than asking an LLM. Essential for content pipelines."},{"id":39,"name":"Grounding DINO","tier":"best","isDefault":true,"priceSats":5,"description":"Open-set object detection using Grounding DINO. Detects any object from text description — returns bounding boxes with coordinates and confidence scores. Structured output agents can parse directly."},{"id":40,"name":"Bria Eraser","tier":"best","isDefault":true,"priceSats":130,"description":"Remove unwanted objects from images using Bria Eraser. Provide image + binary mask (white=erase, black=keep). Neural inpainting fills removed regions convincingly. Trained on licensed data (commercially safe)."},{"id":41,"name":"DDColor","tier":"best","isDefault":true,"priceSats":5,"description":"Colorize black-and-white photos using DDColor. Dual-decoder architecture produces vivid, realistic colors. Impossible for text/vision LLMs to generate color pixel data."},{"id":42,"name":"NAFNet","tier":"best","isDefault":true,"priceSats":110,"description":"Fix blurry or out-of-focus photos using NAFNet. Neural pixel restoration that recovers actual detail — not just sharpening. Also supports denoising."},{"id":48,"name":"Depth Anything V2","tier":"best","isDefault":true,"priceSats":5,"description":"Estimate per-pixel depth from a single photo. Depth Anything V2 (NeurIPS 2024) returns both the raw greyscale depth map for 3D, AR and ControlNet pipelines and a colour visualization. Structured geometry a vision LLM cannot produce."}],"defaultModelId":48},"Fax":{"models":[{"id":44,"name":null,"tier":null,"isDefault":false,"priceSats":500}],"defaultModelId":null},"FaxInbound":{"models":[{"id":45,"name":null,"tier":null,"isDefault":false,"priceSats":500}],"defaultModelId":null}},"fixedServices":{"sms":{"description":"Send SMS to any phone number worldwide","pricingNote":"dynamic — varies by destination country"},"call":{"description":"Place an automated phone call with text-to-speech or audio URL","pricingNote":"dynamic — varies by destination country and call duration"},"email":{"description":"Send email with optional attachments","pricingNote":"200 sats per email"},"file_convert":{"description":"Convert files between formats (PDF, DOCX, PNG, etc.)","pricingNote":"100 sats per conversion"},"pdf_merge":{"description":"Merge multiple PDFs into one","pricingNote":"100 sats per merge"},"html_to_pdf":{"description":"Convert HTML or Markdown to PDF","pricingNote":"50 sats per conversion"},"epub_audiobook":{"description":"Convert EPUB books to full AI-narrated audiobooks (zip of chapter MP3s)","pricingNote":"dynamic — ~1 sat per 33 characters, minimum 500 sats"},"ai_call":{"description":"AI-powered phone call with a custom agent persona","pricingNote":"dynamic — telephony + AI cost per minute, minimum 50 sats"},"translate":{"description":"Translate text between languages (100+ languages)","pricingNote":"dynamic — 0.001 sats per character"},"remove_background":{"description":"Remove background from images","pricingNote":"flat rate per image"},"upscale_image":{"description":"Upscale image resolution","pricingNote":"flat rate per image"},"restore_face":{"description":"Restore and enhance faces in photos","pricingNote":"flat rate per image"},"detect_nsfw":{"description":"Detect NSFW content in images","pricingNote":"flat rate per image"},"detect_objects":{"description":"Detect and label objects in images","pricingNote":"flat rate per image"},"remove_object":{"description":"Remove objects from images using a mask","pricingNote":"flat rate per image"},"colorize_image":{"description":"Colorize black-and-white images","pricingNote":"flat rate per image"},"deblur_image":{"description":"Deblur and sharpen images","pricingNote":"flat rate per image"}},"resolution":{"description":"How model selection works for each API call","strategies":["1. Omit modelId → default (best) model is used automatically","2. Pass tier='best'|'better'|'standard' → best model in that tier","3. Pass modelId=<number> → exact model pinning"],"note":"Models may be upgraded behind stable IDs. Pin by ID for reproducibility, or omit for automatic best-model routing."},"_links":{"list_models":"GET /api/l402/models","mcp_discovery":"GET /api/mcp/discovery","l402_docs":"https://sats4ai.com/l402"}}