{"models":[{"id":2,"category":"image","tier":"standard","isDefault":false,"status":"active","name":"Standard","agentDescription":"Fast image generation (xAI Grok Imagine 2). 100 sats flat, 1K. Lowest cost option for simple prompts.","priceSats":100},{"id":3,"category":"audio","tier":null,"isDefault":false,"status":"active","name":"Minimax Speech-2.8-HD","agentDescription":"Convert text to natural speech audio. Multiple voice options. 300 sats flat. Returns MP3.","priceSats":1},{"id":5,"category":"vision","tier":null,"isDefault":true,"status":"active","name":"Vision Chat","agentDescription":"Analyze and describe image content. Ask questions about any image. 21 sats flat. Send image as base64.","priceSats":21},{"id":6,"category":"text","tier":"best","isDefault":true,"status":"active","name":"Kimi K3","agentDescription":"Best quality text model. 262K context window. Supports vision via imageBase64 param. Dynamic pricing: 100 chars/sat, minimum 15 sats.","priceSats":1},{"id":8,"category":"image","tier":"better","isDefault":false,"status":"active","name":"Better","agentDescription":"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.","priceSats":150},{"id":9,"category":"Test","tier":null,"isDefault":false,"status":"active","name":"Contact","agentDescription":null,"priceSats":1},{"id":10,"category":"audio","tier":"best","isDefault":true,"status":"active","name":"Transcription","agentDescription":"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.","priceSats":10},{"id":11,"category":"file","tier":null,"isDefault":false,"status":"active","name":null,"agentDescription":null,"priceSats":100},{"id":13,"category":"file","tier":null,"isDefault":false,"status":"active","name":null,"agentDescription":null,"priceSats":100},{"id":15,"category":"SearchAI","tier":null,"isDefault":false,"status":"active","name":"Search","agentDescription":null,"priceSats":21},{"id":16,"category":"image","tier":"best","isDefault":true,"status":"active","name":"Top-ranked","agentDescription":"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.","priceSats":1200},{"id":17,"category":"3d","tier":"best","isDefault":true,"status":"active","name":"Best","agentDescription":"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.","priceSats":1600},{"id":18,"category":"image","tier":null,"isDefault":false,"status":"active","name":"Best","agentDescription":null,"priceSats":200},{"id":20,"category":"communication","tier":null,"isDefault":false,"status":"active","name":null,"agentDescription":null,"priceSats":null,"pricing":"dynamic","pricingNote":"Destination-dependent. POST without Authorization to receive a priced invoice."},{"id":22,"category":"VoiceCloning","tier":null,"isDefault":false,"status":"active","name":"Voice Cloning","agentDescription":null,"priceSats":7500},{"id":23,"category":"audio","tier":"best","isDefault":true,"status":"active","name":"Music-3","agentDescription":"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.","priceSats":75},{"id":24,"category":"video","tier":"best","isDefault":true,"status":"active","name":"Image to Video","agentDescription":"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.","priceSats":100},{"id":25,"category":"video","tier":"best","isDefault":true,"status":"active","name":"Video Generation","agentDescription":"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.","priceSats":50},{"id":26,"category":"file","tier":null,"isDefault":false,"status":"active","name":null,"agentDescription":null,"priceSats":100},{"id":27,"category":"document","tier":"best","isDefault":true,"status":"active","name":"Mistral OCR","agentDescription":"Extract text from PDFs and images as clean Markdown. 10 sats per page. Send document as base64.","priceSats":10},{"id":28,"category":"file","tier":null,"isDefault":false,"status":"active","name":null,"agentDescription":null,"priceSats":200},{"id":29,"category":"file","tier":"standard","isDefault":true,"status":"active","name":"HTML to PDF","agentDescription":null,"priceSats":50},{"id":30,"category":"file","tier":"standard","isDefault":true,"status":"active","name":"Receipt OCR","agentDescription":null,"priceSats":25},{"id":31,"category":"text","tier":"standard","isDefault":false,"status":"active","name":"Standard","agentDescription":"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.","priceSats":1},{"id":33,"category":"audio","tier":"best","isDefault":true,"status":"active","name":"Inworld TTS 1.5 Max","agentDescription":null,"priceSats":1},{"id":35,"category":"ImageProcess","tier":"best","isDefault":true,"status":"active","name":"BiRefNet","agentDescription":"Background removal using BiRefNet (state-of-the-art, Papers with Code). Handles hair, fur, glass, transparency, complex edges. Returns transparent PNG.","priceSats":44},{"id":36,"category":"ImageProcess","tier":"best","isDefault":true,"status":"active","name":"Real-ESRGAN","agentDescription":"Image upscaling 2x-4x using Real-ESRGAN (ICCV 2021, 100M+ production runs). Neural super-resolution that recovers real detail. Optional face enhancement.","priceSats":5},{"id":37,"category":"ImageProcess","tier":"best","isDefault":true,"status":"active","name":"CodeFormer","agentDescription":"Face restoration using CodeFormer (NeurIPS 2022, state-of-the-art on CelebA-Test). Restores blurry, damaged, or AI-generated faces. Adjustable fidelity parameter.","priceSats":25},{"id":38,"category":"ImageProcess","tier":"best","isDefault":true,"status":"active","name":"NSFW Detector","agentDescription":"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.","priceSats":2},{"id":39,"category":"ImageProcess","tier":"best","isDefault":true,"status":"active","name":"Grounding DINO","agentDescription":"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.","priceSats":5},{"id":40,"category":"ImageProcess","tier":"best","isDefault":true,"status":"active","name":"Bria Eraser","agentDescription":"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).","priceSats":130},{"id":41,"category":"ImageProcess","tier":"best","isDefault":true,"status":"active","name":"DDColor","agentDescription":"Colorize black-and-white photos using DDColor. Dual-decoder architecture produces vivid, realistic colors. Impossible for text/vision LLMs to generate color pixel data.","priceSats":5},{"id":42,"category":"ImageProcess","tier":"best","isDefault":true,"status":"active","name":"NAFNet","agentDescription":"Fix blurry or out-of-focus photos using NAFNet. Neural pixel restoration that recovers actual detail — not just sharpening. Also supports denoising.","priceSats":110},{"id":43,"category":"audio","tier":null,"isDefault":false,"status":"active","name":"OmniVoice TTS","agentDescription":null,"priceSats":1},{"id":44,"category":"Fax","tier":null,"isDefault":false,"status":"active","name":null,"agentDescription":null,"priceSats":500},{"id":45,"category":"FaxInbound","tier":null,"isDefault":false,"status":"active","name":null,"agentDescription":null,"priceSats":500},{"id":46,"category":"file","tier":null,"isDefault":false,"status":"active","name":"E-Signature","agentDescription":null,"priceSats":1000},{"id":47,"category":"file","tier":null,"isDefault":false,"status":"active","name":"Boarding Pass Wallet","agentDescription":null,"priceSats":100},{"id":48,"category":"ImageProcess","tier":"best","isDefault":true,"status":"active","name":"Depth Anything V2","agentDescription":"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.","priceSats":5},{"id":53,"category":"file","tier":"standard","isDefault":false,"status":"active","name":"Render Card","agentDescription":"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.","priceSats":5}],"usage":"Use modelId (number) or tier alias ('best', 'standard') in L402 requests. Omit model param for default."}