chore(model gallery): 🤖 add 1 new models via gallery agent

gallery/index.yaml

@@ -3127,7 +3127,7 @@
     - gemma3
     - gemma-3
   overrides:
     #mmproj: gemma-3-27b-it-mmproj-f16.gguf
     parameters:
       model: gemma-3-27b-it-Q4_K_M.gguf
   files:
@@ -3145,7 +3145,7 @@
   description: |
     google/gemma-3-12b-it is an open-source, state-of-the-art, lightweight, multimodal model built from the same research and technology used to create the Gemini models. It is capable of handling text and image input and generating text output. It has a large context window of 128K tokens and supports over 140 languages. The 12B variant has been fine-tuned using the instruction-tuning approach. Gemma 3 models are suitable for a variety of text generation and image understanding tasks, including question answering, summarization, and reasoning. Their relatively small size makes them deployable in environments with limited resources such as laptops, desktops, or your own cloud infrastructure.
   overrides:
     #mmproj: gemma-3-12b-it-mmproj-f16.gguf
     parameters:
       model: gemma-3-12b-it-Q4_K_M.gguf
   files:
@@ -3163,7 +3163,7 @@
   description: |
     Gemma is a family of lightweight, state-of-the-art open models from Google, built from the same research and technology used to create the Gemini models. Gemma 3 models are multimodal, handling text and image input and generating text output, with open weights for both pre-trained variants and instruction-tuned variants. Gemma 3 has a large, 128K context window, multilingual support in over 140 languages, and is available in more sizes than previous versions. Gemma 3 models are well-suited for a variety of text generation and image understanding tasks, including question answering, summarization, and reasoning. Their relatively small size makes it possible to deploy them in environments with limited resources such as laptops, desktops or your own cloud infrastructure, democratizing access to state of the art AI models and helping foster innovation for everyone. Gemma-3-4b-it is a 4 billion parameter model.
   overrides:
     #mmproj: gemma-3-4b-it-mmproj-f16.gguf
     parameters:
       model: gemma-3-4b-it-Q4_K_M.gguf
   files:
@@ -6781,7 +6781,7 @@
       sha256: 2756551de7d8ff7093c2c5eec1cd00f1868bc128433af53f5a8d434091d4eb5a
       uri: huggingface://Triangle104/Nano_Imp_1B-Q8_0-GGUF/nano_imp_1b-q8_0.gguf
 - &qwen25
   name: "qwen2.5-14b-instruct" ## Qwen2.5
   icon: https://avatars.githubusercontent.com/u/141221163
   url: "github:mudler/LocalAI/gallery/chatml.yaml@master"
   license: apache-2.0
@@ -22237,3 +22237,54 @@
     - filename: Logics-Qwen3-Math-4B.Q4_K_M.gguf
       sha256: 05528937a4cb05f5e8185e4e6bc5cb6f576f364c5482a4d9ee6a91302440ed07
       uri: huggingface://mradermacher/Logics-Qwen3-Math-4B-GGUF/Logics-Qwen3-Math-4B.Q4_K_M.gguf
+- !!merge <<: *qwen3
+  name: "qwen3-next-80b-a3b-instruct"
+  urls:
+    - https://huggingface.co/lefromage/Qwen3-Next-80B-A3B-Instruct-GGUF
+  description: |
+    **Model Name:** Qwen3-Next-80B-A3B-Instruct
+    **Author:** Qwen (Alibaba Cloud)
+    **License:** Apache 2.0
+
+    ### 📌 Overview
+    Qwen3-Next-80B-A3B-Instruct is a highly efficient, ultra-long context, instruction-tuned large language model based on the Qwen3-Next architecture. It achieves strong performance with only 3 billion activated parameters (out of 80B total), thanks to a hybrid attention mechanism and high-sparsity Mixture-of-Experts (MoE).
+
+    ### 🔍 Key Features
+    - **Model Type:** Causal Language Model (Instruct)
+    - **Parameters:** 80B total | 3B activated (MoE)
+    - **Context Length:** Up to **262,144 tokens** natively, extendable to **1,010,000 tokens** using YaRN RoPE scaling
+    - **Architecture:** Hybrid Attention (Gated DeltaNet + Gated Attention), MoE with 512 experts (10 active per layer), stability-optimized normalization
+    - **Training:** 15 trillion tokens (pretraining), followed by post-training
+    - **Use Case:** Ideal for long-form content generation, complex reasoning, coding, and agentic tasks requiring extended context
+
+    ### ✅ Performance Highlights
+    - Matches or exceeds larger models like Qwen3-235B-A22B-Instruct on key benchmarks
+    - Superior inference speed and efficiency—10x throughput over Qwen3-32B on long contexts
+    - Outstanding results on MMLU-Pro (80.6), LiveBench (75.8), and coding tasks (LiveCodeBench 56.6)
+
+    ### 🛠️ Deployment & Use
+    - **Framework Support:** Hugging Face Transformers, vLLM, SGLang
+    - **Recommended Inference:** Use vLLM or SGLang with MTP (Multi-Token Prediction) for maximum speed
+    - **Ultra-Long Context:** Enable YaRN scaling for inputs exceeding 256K tokens
+
+    ### 📚 Citation
+    ```bibtex
+    @misc{qwen3technicalreport,
+      title={Qwen3 Technical Report},
+      author={Qwen Team},
+      year={2025},
+      eprint={2505.09388},
+      archivePrefix={arXiv},
+      primaryClass={cs.CL},
+      url={https://arxiv.org/abs/2505.09388}
+    }
+    ```
+
+    > 🔗 **Try it now**: [Qwen Chat](https://chat.qwen.ai/) or deploy via [Hugging Face](https://huggingface.co/Qwen/Qwen3-Next-80B-A3B-Instruct)
+  overrides:
+    parameters:
+      model: Qwen__Qwen3-Next-80B-A3B-Instruct-Q4_K_M.gguf
+  files:
+    - filename: Qwen__Qwen3-Next-80B-A3B-Instruct-Q4_K_M.gguf
+      sha256: d16cdbe3d1aa2427862f41ebce219b81cc3128a585c29d6f60c3daaf40a05dd3
+      uri: huggingface://lefromage/Qwen3-Next-80B-A3B-Instruct-GGUF/Qwen__Qwen3-Next-80B-A3B-Instruct-Q4_K_M.gguf