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

gallery/index.yaml

@@ -3309,7 +3309,7 @@
     - gemma3
     - gemma-3
   overrides:
     #mmproj: gemma-3-27b-it-mmproj-f16.gguf
     parameters:
       model: gemma-3-27b-it-Q4_K_M.gguf
   files:
@@ -3327,7 +3327,7 @@
   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:
@@ -3345,7 +3345,7 @@
   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:
@@ -6963,7 +6963,7 @@
       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
@@ -23363,3 +23363,27 @@
     - filename: Qwen3-Grand-Horror-Light-1.7B.Q4_K_M.gguf
       sha256: cbbb0c5f6874130a8ae253377fdc7ad25fa2c1e9bb45f1aaad88db853ef985dc
       uri: huggingface://mradermacher/Qwen3-Grand-Horror-Light-1.7B-GGUF/Qwen3-Grand-Horror-Light-1.7B.Q4_K_M.gguf
+- !!merge <<: *qwen3vl
+  name: "gelato-30b-a3b"
+  urls:
+    - https://huggingface.co/noctrex/Gelato-30B-A3B-GGUF
+  description: |
+    **Gelato-30B-A3B** is a state-of-the-art vision-language model designed specifically for GUI (graphical user interface) computer-use tasks. Built on the **Qwen3-VL-30B-A3B-Instruct** foundation, it has been fine-tuned on the large-scale **Click-100k** dataset to achieve expert-level accuracy in understanding and interacting with desktop interfaces.
+
+    Key highlights:
+    - **Strong grounding performance**: Achieves **63.88% accuracy on ScreenSpot-Pro** and **73.40% on OS-World-G**, outperforming prior specialized models like GTA1-32B and even much larger VLMs such as Qwen3-VL-235B.
+    - **Efficient activation**: Only 3.3B parameters activated during inference, making it highly efficient despite its 30B total size.
+    - **Open-source & accessible**: Fully open-source under the Apache 2.0 license, with full training code and datasets available.
+    - **Agent-ready**: When paired with powerful reasoning models like GPT-5, Gelato enables frontier-level agentic behavior for tasks such as web navigation, software interaction, and UI automation.
+
+    Ideal for researchers and developers building AI agents that interact with graphical interfaces in real-world environments.
+    **Repository**: [mlfoundations-cua-dev/Gelato-30B-A3B](https://huggingface.co/mlfoundations-cua-dev/Gelato-30B-A3B)
+    **Dataset**: [Click-100k](https://huggingface.co/datasets/mlfoundations/clicks-100k)
+    **Benchmark Results**: [OS-World Leaderboard](https://github.com/mlfoundations/grounding-model-os-world)
+  overrides:
+    parameters:
+      model: Gelato-30B-A3B-Q4_K_M.gguf
+  files:
+    - filename: Gelato-30B-A3B-Q4_K_M.gguf
+      sha256: d44af330b10adaa291dced867ae90bbbdc8d502f97d994d7209828efe6343824
+      uri: huggingface://noctrex/Gelato-30B-A3B-GGUF/Gelato-30B-A3B-Q4_K_M.gguf