Category

L2P: Unlocking Latent Potential for Pixel Generation

Transforms pretrained latent-diffusion priors into pixel-space diffusion models by removing the VAE and training shallow pixel layers on LDM-generated synthetic images — enabling fast convergence, native 4K output, and low-data training on 8 GPUs.

ZAYA1-8B

Mixture-of-Experts LLM tuned for mathematical and coding reasoning, with ~760M active / 8.4B total parameters and post-training for improved stepwise reasoning. Optimized for inference efficiency (vLLM/transformers forks) so it can run in computation-constrained or local deployments; Apache-2.0 licensed.

havenoammo/Qwen3.6-27B-MTP-UD-GGUF

Merges Unsloth UD XL quantized GGUF of Qwen3.6-27B with compact Q8_0 MTP heads to enable multi-token (speculative) decoding on llama.cpp builds that support MTP; aimed at image-text-to-text usage with reduced MTP overhead.

Supertone/supertonic-3

Converts text into natural-sounding speech locally using compact ONNX TTS assets. Optimized for CPU/edge inference (~99M params) with support for 31 languages, expression tags (e.g., <laugh>), and improved stability versus Supertonic 2 — suitable for on-device multilingual TTS.

Qwopus3.6-35B-A3B-v1

A reasoning-enhanced Mixture-of-Experts (MoE) LLM fine-tuned for multimodal image-text-to-text tasks and long-context reasoning; built on Qwen3.6-35B-A3B with LoRA and released as an experimental GGUF community model.

HiDream-O1-Image-Dev

Distilled dev checkpoint of an image foundation model that natively unifies raw pixels and text tokens for text-to-image, image editing, long-text rendering, and subject-driven personalization at up to 2048×2048. The Dev variant targets faster (28-step) inference for iterative use and research.

Command A+ (command-a-plus-05-2026-bf16)

Processes text and images to produce conversational, reasoning-focused multilingual outputs for agentic workflows. Built as a sparse MoE decoder (25B active / 218B total parameters) with 128K context and available in BF16/FP8/W4A4 quantizations to balance quality and deployability.

Hy-MT2-30B-A3B

A 30B mixture-of-experts multilingual translation model supporting 33 languages and instruction-following translation. Offers MoE architecture, fast-thinking mode, and quantized/deployment-ready variants for production translation and subtitle tasks.

Tencent Hy-MT2-1.8B

A family of multilingual translation models optimized for real-world, instruction-following translation across 33 languages. The 1.8B model targets on-device use with extreme quantization (≈440 MB via AngelSlim), while 7B/30B variants trade size for higher accuracy.

Hy-MT2-7B

7B multilingual translation model optimized for instruction-following and low-latency deployment across 33 languages; provides quantized/FP8/GGUF builds and integrations (vLLM, llama.cpp) for server and on-device inference.

unsloth/Qwen3.6-27B-MTP-GGUF

Provides a quantized GGUF build of Qwen3.6‑27B with MTP (multi‑token prediction) support for faster local inference. Packaged for GGUF-compatible runners (llama.cpp) and Hugging Face/transformers workflows, with deployment notes for CPU/GPU and vLLM/SGLang integration.

Qwen3.6-35B-A3B-MTP-GGUF (unsloth)

A GGUF-quantized build of Qwen3.6-35B packaged by unsloth for local and accelerated inference. Adds MTP speculative decoding guidance and deployment notes for llama.cpp, vLLM, SGLang and long-context/multimodal use cases.