prithivMLmods/Vortex-9B-CodeCore-Merge

1---2license: apache-2.03pipeline_tag: text-generation4library_name: transformers5base_model:6- Qwen/Qwen3.5-9B7- TokenRhythm/NeoHorse-1-9B8- Jackrong/Qwopus3.5-9B-Coder9- ornith-ai/Ornith-1.5-9B10tags:11- text-generation-inference12- pytorch13- omnimergekit14- merge15- qwen3_516- reasoning17- chain-of-thought18- sft19- agent20- tool-use21- function-calling22- coder23- Qwen3.5-9B24language:25- en26---27 28# **Vortex-9B-CodeCore-Merge**29 30> Vortex-9B-CodeCore-Merge is a merged 9B-parameter coding and reasoning model designed for long-horizon coding tasks, agentic coding, and agentic reasoning. It is built by merging [Qwen3.5-9B](https://huggingface.co/Qwen/Qwen3.5-9B) as the base model with [NeoHorse-1-9B](https://huggingface.co/TokenRhythm/NeoHorse-1-9B), [Qwopus3.5-9B-Coder](https://huggingface.co/Jackrong/Qwopus3.5-9B-Coder), and [Ornith-1.5-9B](https://huggingface.co/ornith-ai/Ornith-1.5-9B), combining their capabilities for code generation, multi-step problem solving, instruction following, and autonomous coding workflows. The model is intended for complex software-engineering tasks that require sustained reasoning across multiple steps, code understanding, modification, debugging, and tool-oriented agentic workflows. This model is experimental and may generate artifacts.31 32- *GGUF: [Vortex-9B-CodeCore-Merge-GGUF](https://huggingface.co/prithivMLmods/Vortex-9B-CodeCore-Merge-GGUF). Note: The Multi-Token Prediction (MTP) heads are not preserved in this format. The model runs as a standard single-token-per-step autoregressive decoder.*33 34## Merge Recipe35 36```bash37hf download TokenRhythm/NeoHorse-1-9B \38  --local-dir ./models/NeoHorse-1-9B39 40hf download Jackrong/Qwopus3.5-9B-Coder \41  --local-dir ./models/Qwopus3.5-9B-Coder42 43hf download ornith-ai/Ornith-1.5-9B \44  --local-dir ./models/Ornith-1.5-9B45 46hf download Qwen/Qwen3.5-9B \47  --local-dir ./models/Qwen3.5-9B48```49 50```bash51python omnimergekit.py \52  --base      models/Qwen3.5-9B --task-base ./models/Qwen3.5-9B \53  --source    models/NeoHorse-1-9B \54  --source    models/Qwopus3.5-9B-Coder \55  --source    models/Ornith-1.5-9B \56  --weights   0.55,0.30,0.15 \57  --method omnimerge_v2 --density 0.53 --darex-q 0.75 --seed 42 \58  --no-auto-mlp-skip --skip-patterns visual.,mtp. \59  --output prithivMLmods/Vortex-9B-CodeCore-Merge60```61 62## omnimergekit63 64Tools for model merging, expert pruning, differential competence-map extraction, and GGUF quantization — [omnimergekit](https://github.com/mann1x/omnimergekit)