Alibaba's Qwen team has open-sourced Qwen3.6-35B-A3B, a sparse mixture-of-experts (MoE) model designed for high performance with low computational costs. While the model possesses 35 billion total parameters, it only activates 3 billion during operation, allowing it to outperform larger dense models in logical reasoning and programming tasks.
Key highlights:
- Uses MoE architecture to achieve high intelligence with minimal activated parameters.
- Demonstrates exceptional multimodal capabilities, particularly in spatial intelligence and visual perception.
- Competes closely with large-scale models like Gemma4-31B and Claude Sonnet 4.5 in specific metrics.
- Integrated into Qwen Studio and available via Alibaba Cloud BaiLian as qwen3.6-flash.
- Supports advanced features like thinking chain retention and seamless integration with AI programming assistants.
Key highlights:
- Uses MoE architecture to achieve high intelligence with minimal activated parameters.
- Demonstrates exceptional multimodal capabilities, particularly in spatial intelligence and visual perception.
- Competes closely with large-scale models like Gemma4-31B and Claude Sonnet 4.5 in specific metrics.
- Integrated into Qwen Studio and available via Alibaba Cloud BaiLian as qwen3.6-flash.
- Supports advanced features like thinking chain retention and seamless integration with AI programming assistants.
graphify is an AI coding assistant skill that transforms codebases, documents, and images into a structured, queryable knowledge graph. By utilizing deterministic AST parsing via tree-sitter for code and multimodal LLM capabilities for unstructured data like PDFs and screenshots, it creates a comprehensive map of concepts and relationships. This allows developers to understand complex architectures faster and find the "why" behind design decisions. A key advantage is its massive reduction in token usage per query compared to reading raw files, making it highly efficient for large-scale projects. The tool supports 19 programming languages and integrates seamlessly with platforms like Claude Code and Codex, providing an interactive, persistent, and highly organized way to navigate any codebase or research corpus.
This document details how to run Google's Gemma 4 models locally, including the E2B, E4B, 26B-A4B, and 31B variants. Gemma 4 is a family of open models supporting over 140 languages and up to 256K context, available in both dense and MoE configurations. The E2B and E4B models support image and audio input. These models can be run locally on your device and fine-tuned using Unsloth Studio. The document outlines hardware requirements, recommended settings, and best practices for prompting and multimodal use, including guidance on context length and thinking mode.
This article explains how to implement function calling with Google’s Gemma 3 27B model. It covers the concept of function calling, the step‑by‑step workflow, and provides a practical example using a Python `convert` function to turn $200,000 into EUR. The post walks through prompting Gemma, parsing its `tool_code` output, executing the function with `eval`, and returning a friendly final response. It also demonstrates how to set up the Google‑GenAI SDK, create a chat session, and extract tool calls. The discussion highlights Gemma’s multilingual, multimodal, and agentic capabilities, making it suitable for real‑world AI assistants that need to interact with external APIs and tools.
Microsoft's Phi-4-Reasoning-Vision-15B model challenges the trend of ever-larger AI models by demonstrating strong reasoning capabilities with a comparatively compact size. Trained on curated reasoning data, it aims to achieve performance without the massive compute costs associated with frontier models. The model supports multimodal tasks, combining text and image understanding, and offers flexible reasoning modes for different workloads. This research highlights the importance of data quality and training strategy, suggesting that smarter training techniques can be as impactful as simply increasing model size, particularly for AI agents and practical deployments.
This study introduces a domain-specific Large Vision-Language Model, Human-Scene Vision-Language Model (HumanVLM), designed to provide a foundation for human-scene Vision-Language tasks. They create a large-scale human-scene multimodal image-text dataset (HumanCaption-10M), develop a captioning approach for human-centered images, and train a HumanVLM.
LLMII uses a local LLM to label metadata and index images. It does not rely on a cloud service or database. A visual language model runs on your computer and is used to create captions and keywords for images in a directory tree. The generated information is then added to each image file's metadata.
This post explores how developers can leverage Gemini 2.5 to build sophisticated robotics applications, focusing on semantic scene understanding, spatial reasoning with code generation, and interactive robotics applications using the Live API. It also highlights safety measures and current applications by trusted testers.
Alibaba Cloud released its Qwen2.5-Omni-7B multimodal AI model, designed for cost-effective AI agents and capable of processing various inputs like text, images, audio, and video.
Mistral Small 3.1 is an open-source multimodal AI model optimized for consumer hardware, offering strong performance in text and image processing, multilingual capabilities, and a balance between performance and accessibility. While excelling in many areas, it has limitations in long-context tasks and Middle Eastern language support.
