AI Model Catalogue

120b open-weight language model from OpenAI

A feed-forward neural network that offers a novel approach to visual geometry reconstruction.

A next-generation image generation and editing model from Alibaba's Qwen team. Supports text-to-image and image editing with strong text rendering, especially for Chinese.

Granite-4.0-H-Small is a 32B parameter long-context instruct model finetuned from Granite-4.0-H-Small-Base using a combination of open source instruction datasets with permissive license and internally collected synthetic datasets.

Advanced 20B open-weight reasoning models to customize for any use case and run anywhere.

Compact Powerhouse for Vector Embeddings

Updated Qwen3 model for instruction following

FLUX LoRA trained on ancient Mediterranean paintings — Greek vase paintings, Minoan frescoes, Egyptian tomb murals. Flat schematic figuration, bold outlines, ceremonial character.

Create realistic talking avatar videos from text with HeyGen's Avatar IV engine

FLUX LoRA trained on paintings by J.-A.-D. Ingres. Porcelain-smooth surfaces, exquisitely elongated forms, precise linear contours, neoclassical mastery.

FLUX LoRA trained on 18th century fashion plates. Hand-coloured engraved illustrations, clean white backgrounds, elegant elongated figures, precise draughtsmanship.

DeepSeek-V3-0324 is the leading non-reasoning model, a milestone for open source

Ace Step 1.5 open source music generation model

Private instance of stable-diffusion

Apple's SHARP model — single image to 3D Gaussian splats

FLUX LoRA trained on ancient Roman paintings — Pompeii frescoes, encaustic panel portraits. Earthy terracotta tones, flat modelling, direct frontal gazes.

OpenAI's most capable frontier model for complex professional work, coding, and multi-step reasoning.

FLUX LoRA trained on paintings by Édouard Manet. Loose spontaneous brushwork, flat areas of tone, modern Parisian subjects.

League of Legends splash art generator

FLUX LoRA trained on paintings by Jean-Marc Nattier. Ladies of the court depicted as goddesses, flowing blue drapery, soft pastel tones, Rococo elegance.

Seedream 5.0 lite: image generation with built-in reasoning, example-based editing, and deep domain knowledge

Recraft's latest image generation model at ~2048px resolution. Same design taste and prompt accuracy as V4, with higher resolution for print-ready and large-scale work.

Kling Video 3.0 Omni: Unified multimodal video generation with reference images, video editing, native audio, and multi-shot control

90's Photographs in the style of Bjorks promo material

A feed-forward neural network that offers a novel approach to visual geometry reconstruction.

Remove backgrounds with real alpha transparency using Nano Banana 2. Triangulation matting produces clean edges, proper semi-transparency, and accurate colors — superior to traditional background removal. Optionally specify what to isolate.

Designed to make images sharper and cleaner, Crisp Upscale increases overall quality, making visuals suitable for web use or print-ready materials.

Recraft's latest image generation model, built around design taste. Strong prompt accuracy, art-directed composition, and integrated text rendering. Fast and cost-efficient at standard resolution.

Nano Banana 2 with alpha transparency. Generates images with real RGBA transparency using triangulation matting — clean edges, proper semi-transparency, and accurate colors. Powered by Google Gemini 3.1 Flash Image.

Feed-forward neural network that directly infers all key 3D attributes of a scene.

FLUX LoRA trained on dramatic painted sky imagery — stormy heavens, golden hour light, baroque clouds.

HeartMuLa: A Family of Open Sourced Music Foundation Models

Kimi K2 Thinking is the latest, most capable version of an open-source thinking model.

An extension of AiCoverGen, which provides several new features and improvements, enabling users to generate audio-related content using RVC with ease. Ideal for people who want to incorporate singing functionality into their AI assistant/chatbot/vtuber,

Ideal for rapid ideation and mobile workflows. Perfect for creators who need instant feedback, real-time previews, or high-throughput content.