Brief-details: Llama-4-Maverick is Meta's 17B parameter instruction-tuned language model using FP8 quantization, focused on reliable task completion and following instructions.
Brief Details: A 32B parameter reasoning-focused LLM optimized for roleplay and creative writing, featuring enhanced multi-turn chat capabilities and reduced cross-context repetition.
BRIEF-DETAILS: 32B parameter LLM with multiple GGUF quantizations (9GB-65GB) optimized for different hardware/RAM configurations, featuring high-quality compression and ARM/AVX support
Brief Details: Dream-v0-Instruct-7B is an open-source 7B parameter instruction-tuned LLM focused on diffusion tasks, developed by Dream-org for advanced NLP applications.
Brief-details: AnimeGamer is a groundbreaking AI model for infinite anime life simulation using MLLMs to generate dynamic animation and character states with contextual consistency.
BRIEF DETAILS: Llama-based TTS model with 1B parameters supporting multilingual speech synthesis, one-shot voice cloning, and automatic word alignment across 12+ languages.
Brief Details: Meta's 17B parameter Llama-4 variant optimized for instruction-following tasks with enhanced capabilities and Meta's privacy policy compliance
BRIEF-DETAILS: OpenThinker2-32B is a fine-tuned version of Qwen2.5-32B-Instruct, trained on OpenThoughts2-1M dataset, achieving state-of-the-art performance in math and reasoning tasks.
BRIEF DETAILS: Fine-tuned 12B parameter Gemma model optimized for code generation, showing 32.9% accuracy on LiveCodeBench and trained using TRL framework.
Brief-details: A 27B parameter AI model built on Gemma3, specialized in reasoning, coding, and creative tasks. Features 128K context window, multimodal capabilities, and improved benchmark performance.
BRIEF DETAILS: Advanced Redux adapter for Flex.1-alpha with enhanced 512x512 vision encoding using SigLIP2, offering better image quality and commercial usage under Apache 2.0.
Brief Details: Meta's 17B parameter LLaMA-4 Scout model, optimized for instruction-following with enhanced reasoning and capabilities while adhering to Meta's privacy standards.
Brief Details: Meta's 17B parameter Scout variant of Llama-4, featuring 16 experts and designed for advanced language understanding and generation tasks.
Brief-details: IBM's 8B parameter speech-language model for ASR/AST tasks. Supports English speech recognition and translation to major languages. Built on Granite architecture with conformer blocks.
BRIEF DETAILS: Meta's Llama-4-Maverick-17B-128E: A 17 billion parameter large language model from Meta's Llama family, featuring enhanced capabilities and Meta's privacy-compliant data processing.
Brief-details: Lumina-mGPT-2.0 is an advanced multilingual GPT model developed by Alpha-VLLM, designed for enhanced language understanding and generation across multiple languages.
BRIEF-DETAILS: Gemma 3B 12-bit quantized model from Google, optimized for inference tasks with GGUF format compatibility and Q4 precision
Brief Details: 4-bit quantized version of Meta's Llama 4 Scout (17B parameters, 16 experts) optimized with Unsloth's Dynamic Quants for efficient deployment while maintaining accuracy.
BRIEF-DETAILS: Meta's Llama-4-Maverick-17B-128E-Original is a large language model with 17B parameters, designed for advanced language understanding and generation tasks.
Brief-details: VACE is a 1.3B parameter all-in-one video creation and editing model supporting reference-to-video, video-to-video editing, and masked video editing tasks.
Brief Details: Gemma 3 4B instruction-tuned model with QAT quantization. High-performance open vision-language model from Google supporting text & image inputs with 128K context window.