Qwen Team Open-Sources Qwen3.6-35B-A3B: A Sparse MoE Vision-Language Model with 3B Active Parameters and Agentic Coding Capabilities

We break down Qwen-RobotSuite, the Qwen team's three new embodied AI models. We cover RobotManip, a Vision-Language-Action model built on Qwen3.5-4B for manipulation. We cover RobotWorld, a language-conditioned video world model with a 60-layer MMDiT. We cover RobotNav, a navigation model built on Qwen3-VL across 2B, 4B, and 8B sizes. We walk through the architecture, data pipelines, and benchmark results for each. The post Meet Qwen-RobotSuite: Three Embodied AI Models for VLA Manipulation, Video World Modeling, and Navigation appeared first on MarkTechPost.

Qwen3.7-Plus is Alibaba's multimodal agent model on Bailian, understanding images and video while adding self-programming and tool invocation. The post Alibaba’s Qwen Team Launches Qwen3.7-Plus, Adding Vision, Deep Reasoning, Tool Invocation, and Autonomous Iteration on the Bailian Platform appeared first on MarkTechPost.

StepFun releases Step 3.7 Flash, a 198B MoE model with native vision, 256k context, and Advisor Mode. The post StepFun Releases Step 3.7 Flash: A 198B MoE Vision-Language Model for Coding Agents and Search Workflows appeared first on MarkTechPost.

Alibaba’s Qwen team has unveiled Qwen3.7-Max, a flagship model built for the agent era. Unlike conventional chatbot-focused LLMs, it is designed as a foundation for autonomous AI agents that can code, debug, use tools, manage workflows, and execute long-running enterprise tasks. Alibaba claims the model can operate autonomously for up to 35 hours without performance […] The post Qwen3.7-Max: Alibaba’s New Agent-First LLM for Coding, Reasoning, and Long-Horizon AI Workflows  appeared first on Analytics Vidhya.

In this tutorial, we explore OpenMythos by building an advanced recurrent-depth transformer workflow that runs end-to-end in Google Colab. We create both MLA and GQA model variants, compare their parameter counts, and check the stability of the recurrent injection matrix through its spectral radius. The post Build Recurrent-Depth Transformers with OpenMythos for MLA, GQA, Sparse MoE, and Loop-Scaled Reasoning appeared first on MarkTechPost.

Alibaba's Qwen team introduced Qwen3.7-Max at the 2026 Alibaba Cloud Summit, describing it as its most advanced and comprehensive agent model to date. The model features a 1M-token context window, extended-thinking mode, and is designed for long-horizon tasks including coding, debugging, and multi-step workflow automation. It scored 56.6 on the Artificial Analysis Intelligence Index, ranking fifth overall among proprietary models. The post Qwen Introduces Qwen3.7-Max: A Reasoning Agent Model With a 1M-Token Context Window appeared first on MarkTechPost.

The QwenLM team has released FlashQLA, a new kernel library that dramatically accelerates the forward and backward passes of Gated Delta Network (GDN) Chunked Prefill, targeting both large-scale pretraining and edge-side agentic inference scenarios. The post Qwen Team Releases FlashQLA: a High-Performance Linear Attention Kernel Library That Achieves Up to 3× Speedup on NVIDIA Hopper GPUs appeared first on MarkTechPost.

Alibaba’s Qwen Team has released Qwen3.6-27B, the first dense open-weight model in the Qwen3.6 family — and arguably the most capable 27-billion-parameter model available today for coding agents. It brings substantial improvements in agentic coding, a novel Thinking Preservation mechanism, and a hybrid architecture that blends Gated DeltaNet linear attention with traditional self-attention — all […] The post Alibaba Qwen Team Releases Qwen3.6-27B: A Dense Open-Weight Model Outperforming 397B MoE on Agentic Coding Benchmarks appeared first on MarkTechPost.