EXAONE 3.5 is a collection of instruction-tuned bilingual (English and Korean) generative models ranging from 2.4B to 32B parameters, developed and released by LG AI Research.
Quantized Q4_K_S version of Devstral small 2 24B FP16 model
Quantized Q4_K_S version of Devstral small 24B FP16 model
Qwen3-4B non-thinking mode, named Qwen3-4B-Instruct-2507
Mistral Small 3.1
Mistral Small 3.1 24B quantization from bartowski Q6_K with tool support
fine-tuned version of mistralai/Mistral-Small-24B-Instruct-2501, optimized specifically for creative writing tasks. by **lars1234**
Mistral-Small-3.1-24B-Instruct-2503-Q4_K_M
Arcee-Blitz (24B) is a new Mistral-based 24B model distilled from DeepSeek, available in [F16, q8_0, q6_K, q4_K_S]
ReadyArt/Omega-Darker_The-Final-Directive-24B (i1-Q6_K)
ReadyArt/Broken-Tutu-24B-Transgression-v2.0 (i1-Q6_K)
ReadyArt/Broken-Tutu-24B-Unslop-v2.0 (i1-Q6_K)
EXAONE Deep by LG AI Research.
TroyDoesAI/BlackSheep-24B (i1-Q6_K)
ReadyArt/Broken-Tutu-24B (i1_Q6_K)
26.3.7. Qwen3.5-27B-Q4 Update: This model introduces higher-quality reasoning trajectories across domains such as science, instruction-following, and mathematics.
K2-Horizon-MoVA-36B-A4B is the sparse member of the K2-Horizon family: a Mixture-of-Experts model with Mixture-of-Values attention (MoVA) that stores 36B parameters and runs 4B per token.
https://huggingface.co/akjindal53244/Llama-3.1-Storm-8B-GGUF