Asif Razzaq writes that Alibaba's Qwen team has released Qwen3.8-Flash-Next, an open-weight multimodal Mixture-of-Experts model pairing a 125B backbone with a 51B N-gram embedding table and a 4B multi-token prediction module, totaling 180B on disk but activating only 6B parameters per token. The architecture blends Gated DeltaNet linear-attention layers (three of every four) with Qwen Sparse Attention at micro-block granularity, and is positioned as the architectural preview of Qwen4 in the same role Qwen3-Next played for Qwen3.5. Training cost is reported at roughly one-ninth that of Qwen3.7-Plus.
- Licensed under qwen-community-1.0, not Apache-2.0; verify terms before commercial use.
- FP8 checkpoint is 172.78 GiB; minimum validated config is TP2 on GB300, so self-hosting requires a multi-GPU node.
- The 20M-entry bigram/trigram table at layer 2 can be offloaded to host memory with asynchronous prefetch (NVIDIA only).
- Claude Opus 4.6 (Max) still leads on HLE (40.0 vs. Qwen's 35.9); DeepSeek-V4-Flash-0731 leads NL2Repo-Bench (54.2 vs. 48.1).
- Native context is 262,144 tokens, extensible to 1,000,000 via YaRN.
- Licensed under qwen-community-1.0, not Apache-2.0; verify terms before commercial use.
- FP8 checkpoint is 172.78 GiB; minimum validated config is TP2 on GB300, so self-hosting requires a multi-GPU node.
- The 20M-entry bigram/trigram table at layer 2 can be offloaded to host memory with asynchronous prefetch (NVIDIA only).
- Claude Opus 4.6 (Max) still leads on HLE (40.0 vs. Qwen's 35.9); DeepSeek-V4-Flash-0731 leads NL2Repo-Bench (54.2 vs. 48.1).
- Native context is 262,144 tokens, extensible to 1,000,000 via YaRN.
Alibaba has open-sourced Qwen-UI-Agent, a GUI agent foundation model that operates across mobile, desktop, web, and deep-search environments on real hardware rather than relying on simulation. It achieves top benchmark results: 82.1% on MobileWorld, 79.5% on OSWorld-Verified, and first on WebArena. It also introduces MobileWorld-Real, a 400+ task benchmark on 100+ phones and 150+ apps, with a 92.2% success rate.
- Supports command-line execution alongside standard GUI operations and batches multiple actions into a single decision step to shorten trajectories.
- Built-in safety layer refuses illegal or high-risk requests outright and pauses at sensitive operations (payments, data deletion, privacy grants) for explicit user confirmation.
- Trained via online reinforcement learning on trajectories exceeding 100 steps, paired with adaptive curriculum learning to progressively tackle longer tasks.
- Supports command-line execution alongside standard GUI operations and batches multiple actions into a single decision step to shorten trajectories.
- Built-in safety layer refuses illegal or high-risk requests outright and pauses at sensitive operations (payments, data deletion, privacy grants) for explicit user confirmation.
- Trained via online reinforcement learning on trajectories exceeding 100 steps, paired with adaptive curriculum learning to progressively tackle longer tasks.
Gilbert Pagayon writes that Alibaba'''s Qwen team released Qwen3.8-27B, a 27-billion-parameter open-weight model under Apache 2.0 that can run quantized versions on consumer hardware for coding, research, multimodal tasks and long agent work while keeping data local. The model uses a dense hybrid-attention architecture with flexible reasoning mode and multi-token prediction, and Alibaba'''s benchmarks show it outperforming Claude Opus 4.6 Max on some coding tests but lagging on others. Early hands-on reports praise its capability for its size, though performance varies with quantization, hardware and settings, making the Opus rivalry a signal rather than a verdict.
- Native context is 262,144 tokens, extendable to ~1 million tokens with configuration
- Vision encoder handles images and video natively; scores 91.1 on OmniDocBench 1.5
- Q4_K_M GGUF is ~17.8GB, fitting a 24GB GPU; full BF16 is ~54.7GB
- Multi-token prediction speed gains depend on workload and engine and can be slower in some cases
- Native context is 262,144 tokens, extendable to ~1 million tokens with configuration
- Vision encoder handles images and video natively; scores 91.1 on OmniDocBench 1.5
- Q4_K_M GGUF is ~17.8GB, fitting a 24GB GPU; full BF16 is ~54.7GB
- Multi-token prediction speed gains depend on workload and engine and can be slower in some cases
OpenSandbox is a general-purpose sandbox platform for AI applications offering multi-language SDKs, unified sandbox APIs and Docker/Kubernetes runtimes for use cases like coding agents, GUI agents, evaluation, code execution and RL training. It provides SDKs, CLI and MCP integration, a sandbox protocol for custom runtimes, built-in environments such as command, filesystem and code interpreter, network ingress/egress controls, credential vault injection and strong isolation via gVisor, Kata Containers and Firecracker.
- Release images published to Docker Hub, GHCR and Alibaba Cloud with Cosign signatures and provenance
- SDKs for Python, Java/Kotlin, TypeScript/JavaScript, C#/.NET and Go
- OpenSSF Best Practices badge and CNCF Landscape listing
- Release images published to Docker Hub, GHCR and Alibaba Cloud with Cosign signatures and provenance
- SDKs for Python, Java/Kotlin, TypeScript/JavaScript, C#/.NET and Go
- OpenSSF Best Practices badge and CNCF Landscape listing
Qwen Team writes that Qwen3.8-27B is presented as the most capable generation in the Qwen open-model family so far, built on the Qwen3.5 foundation with substantial gains across coding, professional work, research and long-horizon agentic tasks. The model is a 27B-parameter dense causal language model with a vision encoder, native 262,144-token context extensible to 1,000,000 tokens, flexible thinking control with reasoning_effort and preserve_thinking, and Multi-Token Prediction for faster inference. The Hugging Face page hosts Unsloth's GGUF quantizations and provides install and run instructions for llama.cpp, Ollama, Unsloth Studio, LM Studio and other local apps.
Open Code Review is an AI-powered CLI tool designed for automated, high-precision code reviews. Originally developed as Alibaba Group's internal assistant, the project uses a hybrid architecture that combines deterministic engineering with LLM agents to provide stable and accurate feedback. Unlike general-purpose agents, it employs smart file bundling and fine-grained rule matching to maintain context and prevent issues like position drift or incomplete coverage on large changesets.
Key features:
- AI-driven line-level review comments
- Hybrid architecture combining hard constraints with dynamic decision-making
- Support for various LLM endpoints including OpenAI and Anthropic
- Seamless integration with CI/CD pipelines and coding agents like Claude Code
- Customizable rule sets for specific project requirements
Key features:
- AI-driven line-level review comments
- Hybrid architecture combining hard constraints with dynamic decision-making
- Support for various LLM endpoints including OpenAI and Anthropic
- Seamless integration with CI/CD pipelines and coding agents like Claude Code
- Customizable rule sets for specific project requirements
This guide provides instructions for running Alibaba's Qwen3.6 multimodal hybrid-thinking models locally using Unsloth tools. It covers the 27B and 35B-A3B variants, which support a 256K context window across 201 languages and excel in agentic coding, vision, and chat tasks. The article details hardware requirements for various quantization levels and explains how to leverage Multi Token Prediction (MTP) for significantly faster inference.
Key topics:
- Hardware memory requirements for quantized models
- Faster generation via Multi Token Prediction (MTP)
- Integration with Unsloth Studio, llama.cpp, and MLX
- Preserved thinking mode configurations
Key topics:
- Hardware memory requirements for quantized models
- Faster generation via Multi Token Prediction (MTP)
- Integration with Unsloth Studio, llama.cpp, and MLX
- Preserved thinking mode configurations
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.
CoPaw is a personal AI assistant designed for easy installation and deployment, whether on your local machine or in the cloud. It supports multiple chat applications and offers easily extensible capabilities. Core features include broad channel support (DingTalk, Feishu, QQ, Discord, iMessage, and more), user control over memory and personalization, and built-in skills with the ability to create custom ones.
CoPaw enables various use cases, from social digests and productivity tools to creative writing and research assistance. It's a versatile teammate for your digital life, aiming to be a helpful "co-paw" by your side.
CoPaw enables various use cases, from social digests and productivity tools to creative writing and research assistance. It's a versatile teammate for your digital life, aiming to be a helpful "co-paw" by your side.
Examples for common OpenSandbox use cases. Each subdirectory contains runnable code and documentation. Integrations and sandboxes are available for various tools and services like AI models, desktop environments, and web scraping.
