AI Model Catalog

Meta's largest and most capable open-source language model with 405 billion parameters, offering state-of-the-art performance across reasoning, coding, and multilingual tasks.

A powerful 70B parameter model that balances performance and efficiency, ideal for production deployments requiring high-quality outputs.

A compact yet capable model perfect for edge deployment and resource-constrained environments while maintaining strong performance.

A sparse mixture-of-experts model that matches or outperforms LLaMA 2 70B while being faster and more efficient through its innovative architecture.

Mistral's largest open model with 141B total parameters, offering exceptional performance across all tasks with efficient sparse activation.

Alibaba's flagship open model with exceptional multilingual capabilities, particularly strong in Chinese and Asian languages.

Specialized coding model that excels at code generation, completion, and debugging across multiple programming languages.

Google's open model built on Gemini research, offering strong performance with efficient architecture and safety features.

Microsoft's efficient small language model that punches above its weight class with strong reasoning and coding abilities.

Advanced coding model with exceptional performance on programming tasks, supporting 338 programming languages.

One of the largest open-source models with 180B parameters, trained on 3.5 trillion tokens.

High-performance bilingual model excelling in both English and Chinese tasks.

Efficient 7B model that outperforms larger models through superior architecture.

Fine-tuned LLaMA model trained on user conversations, excelling at dialogue.

Commercially-usable model with strong performance and flexible licensing.

Fine-tuned Mixtral model optimized for instruction following and reasoning.

High-quality fine-tune focused on instruction following and helpfulness.

Aligned Mistral 7B model optimized for helpful, harmless responses.

RLAIF-trained model achieving strong performance through reinforcement learning.

Depth-upscaled model achieving strong performance through innovative architecture.

Reasoning-focused model trained with progressive learning techniques.

Fine-tuned model optimized for STEM and logical reasoning.

Instruction-tuned model with strong creative writing abilities.

Uncensored model trained for helpfulness without alignment restrictions.

Companion-focused model trained for empathetic conversation.

Long-context model optimized for extended conversations and documents.

C-RLFT trained model achieving strong performance with efficient training.

Intel-optimized model for efficient deployment on Intel hardware.

Efficient language model with strong performance for its size.

Efficient model with strong performance from Adept AI.

Previous generation Meta model, still widely used and capable.

Mid-size LLaMA 2 model balancing performance and efficiency.

Compact LLaMA 2 model for efficient deployment.

Large Qwen model with exceptional multilingual capabilities.

Previous generation Qwen model, still highly capable.

Compact Gemma 2 model with strong performance for its size.

Ultra-compact Gemma 2 for extreme edge deployment.

Smallest Phi-3 model optimized for mobile and edge deployment.

Balanced Phi-3 model with good performance and efficiency.

Mid-size Falcon model with strong performance.

Compact Falcon model for efficient deployment.

Compact Yi model with strong bilingual capabilities.

Chinese-focused model with strong performance in Chinese tasks.

Advanced Chinese model with strong reasoning capabilities.

Bilingual conversational model optimized for Chinese and English.

Massively multilingual model trained on 46 languages.

Compact BLOOM model for efficient multilingual deployment.

Fully open language model with complete training data and code.

Suite of models for studying training dynamics and scaling laws.

Early open-source GPT-style model, historically significant.

Large-scale open-source model from EleutherAI.

Open reproduction of LLaMA with fully open training data.

Open reproduction of LLaMA with permissive licensing.

Compact aligned model for helpful, harmless responses.

Ultra-compact model for extreme edge deployment.

Efficient small model optimized for enterprise deployment.

Fully open model with complete training process transparency.

Advanced code model trained on 600+ programming languages with strong performance.

Evol-Instruct trained coding model with exceptional problem-solving abilities.

Fine-tuned CodeLlama optimized for technical question answering and code generation.

OSS-Instruct trained model achieving strong coding performance.

Compact code model optimized for code completion and IDE integration.

The most advanced open-source image generation model, producing high-quality, photorealistic images from text prompts.

Latest generation of Stable Diffusion with improved text rendering, composition, and photorealism.

Next-generation image model from the creators of Stable Diffusion, offering unprecedented quality and prompt adherence.

Popular and widely-used image generation model with extensive community support.

Improved version with better composition and detail.

Russian-developed image model with unique artistic style.

Aesthetic-focused model optimized for beautiful, artistic outputs.

Efficient high-resolution image generation with fast inference.

Extremely efficient image model with novel compression approach.

Modular cascaded diffusion model with exceptional text rendering.

Three-stage cascaded diffusion model with exceptional quality and efficiency.

Distilled SDXL model for real-time image generation with single-step inference.

Latent Consistency Model adapter for fast few-step generation.

State-of-the-art speech recognition model supporting 99 languages with exceptional accuracy.

Balanced speech recognition model offering good accuracy with reasonable resource usage.

Compact speech recognition for edge deployment and real-time applications.

Text-to-audio model generating speech, music, and sound effects.

Controllable music generation model creating high-quality audio from text.

Suite of audio generation models for music, sound effects, and compression.

Massively multilingual and multimodal translation model.

Ultra-compact speech recognition for extreme edge deployment.

Balanced speech recognition model for general use.

High-quality text-to-speech with voice cloning capabilities.

Multilingual text-to-speech with voice cloning in 17 languages.

Stable Diffusion fine-tune for music generation via spectrograms.

Vision-language model combining visual understanding with language generation.

Powerful vision-language model with strong visual grounding.

Multilingual vision-language model with strong Chinese support.

Efficient vision-language model with strong zero-shot capabilities.

Instruction-tuned vision-language model for diverse visual tasks.

Next generation LLaVA with improved visual reasoning.

Lightweight vision-language model with strong image understanding.

High-quality embedding model for semantic search and retrieval.

Text embedding model with strong performance on retrieval tasks.

Multi-lingual, multi-functionality, multi-granularity embedding model.

General text embedding model with strong performance.

Instruction-based embedding model for customizable representations.