Alibaba's Open-source Qwen3 Omni Model

Today’s top AI Highlights:

1. Alibaba Qwen releases open-source Omni model

2. Unified MCP server that handles 1000s of tools progressively

3. DeepSeek open-sourced V3.1-Terminus model

4. Google’s Windows app can search your files, apps, and the web

5. Build MCP agents with reasoning, system prompts, and orchestration

& so much more!

Read time: 3 mins

AI Tutorial

Learn OpenAI Agents SDK from zero to production-ready!

We have created a comprehensive crash course that takes you through 11 hands-on tutorials covering everything from basic agent creation to advanced multi-agent workflows using OpenAI Agents SDK.

What you'll learn and build:

• Starter agents with structured outputs using Pydantic

• Tool-integrated agents with custom functions and built-in capabilities

• Multi-agent systems with handoffs and delegation

• Production-ready agents with tracing, guardrails, and sessions

• Voice agents with real-time conversation capabilities

Each tutorial includes working code, interactive web interfaces, and real-world examples.

The course covers the complete agent development lifecycle: orchestration, tool integration, memory management, and deployment strategies.

Everything is 100% open-source.

We share hands-on tutorials like this every week, designed to help you stay ahead in the world of AI. If you're serious about leveling up your AI skills and staying ahead of the curve, subscribe now and be the first to access our latest tutorials.

Don’t forget to share this newsletter on your social channels and tag Unwind AI (X, LinkedIn, Threads) to support us!

Latest Developments

Alibaba Qwen Releases Open-Source Omni Model 💬🎤🌅

China’s Alibaba Qwen team keeps shipping the best open-source models and this time it is an omni-modal model, similar to GPT-4o and Gemini’s version that power their Voice Modes.

Qwen3-Omni is a natively end-to-end multilingual omni model that can process text, images, audio, and video, and deliver real-time streaming responses in both text and natural speech.

It's built from the ground up using a novel Thinker-Talker architecture where one component handles reasoning while another generates real-time speech. The model's efficiency is staggering, using only 3B active parameters from 30B total, competing with much larger models on everything from complex reasoning to music analysis. With Apache 2.0 licensing and complete model weights available, you can now build "Her"-style AI assistants without being locked into expensive cloud services.

Key Highlights:

1. Speech Architecture - Uses advanced audio encoding that drives latency down to an industry-leading 211ms.

2. Extended Understanding - Can process up to 30-minute audio sequences while maintaining coherent understanding and generating contextually relevant responses throughout the entire duration.

3. Specialized Audio Captioner - Ships with Qwen3-Omni-30B-A3B-Captioner, an open-source model specifically fine-tuned for detailed, low-hallucination audio descriptions, filling a critical gap in the open-source ecosystem.

4. Customization - Qwen3-Omni can be freely adapted via system prompts to modify response styles, personas, and behavioral attributes.

5. Deployment - Includes Docker containers, vLLM support, and API access through DashScope, plus comprehensive cookbooks covering everything from music analysis to real-time video navigation.

Unified MCP Server that Handles 1000s of Tools Progressively🪜 🧰

AI agents choke when handed too many tools at once. The reasons we all know now - context window overload, tool naming issues, unclear tool prompts, and more. Context overload kills performance, forcing most AI agents to artificially cap themselves at 40-50 tools just to stay functional.

Strata is an open-source unified MCP server by Klavis AI that guides AI agents through 1000s of tools in multiple apps progressively instead of overwhelming them with everything at once.

Strata intelligently pre-loads the integrations that you or your users have already enabled. When an agent needs to accomplish a complex task, Strata works through four intelligent stages: identifying tool categories > available tools > actions to use > and execution - much like a human would identify and use tools.

The results show measurable wins where AI agents can carry out multi-step actions more reliably.

Key Highlights:

1. Cognitive Load Reduction - Mirrors human problem-solving behavior by revealing tool complexity progressively rather than overwhelming agents with complete specifications upfront.

2. Zero-Config Integration - Any external MCP server automatically transforms into discovery-driven architecture without requiring modifications to existing implementations.

3. Built-in Auth Handling - Automatically detects OAuth vs API key authentication, providing white-labeled auth links or guided setup processes when credentials fail.

4. Access Options - Deploy via one-click UI in Klavis dashboard, integrate through API/SDK, or self-host the open-source version for complete control.

Your Shopify DTC Brand Can’t Afford Q4 Without Zipchat

BFCM traffic costs a fortune. If your Shopify brand isn’t converting at its possible best, you’re not just losing sales — you’re burning money and shrinking Q4 margins.

Zipchat.ai is the AI Agent built for DTC ecommerce. It doesn’t just chat — it sells.

• Closes hesitant shoppers instantly with product answers and recommendations

• Recovers abandoned carts automatically via web + WhatsApp

• Automates support 24/7 so you scale without extra headcount

• Boosts profit margins in Q4, when every order counts

That’s why brands like Police, TropicFeel, and Jackery — brands with 10k visitors/month to millions — trust Zipchat to handle their busiest quarter and fully embrace Agentic Commerce.

Setup takes less than 20 minutes with our success manager. And you’re fully covered with 37 days risk-free (7-day free trial + 30-day money-back guarantee).

On top, use the NEWSLETTER10 coupon for 10% off forever.

Try Zipchat free

Quick Bites

Perplexity releases a personal AI assistant for your inbox
Perplexity just introduced its Email Assistant for Gmail and Outlook, giving Max subscribers a personal AI inside their inbox. It automatically organizes threads, labels priorities, and even prepares daily summaries so you don’t lose track of important updates. Calendar meetings by cc’ing your assistant on any email, and it’ll handle the back-and-forth with your contact. You can also ask it questions about your inbox: "What emails should I prioritize before my board meeting?" "Summarize all messages about the Q4 budget.” It even drafts replies in your own style

Google’s Windows app can search your files, apps, Drive, and the web
Google just launched a Windows app experiment in Labs that lets you search your PC, Google Drive, installed apps, and the web - all without leaving your current window. Hit Alt + Space anytime, even during a game or while writing, to pull up results fast. There’s also a built-in Lens: highlight anything on your screen (text, image) to translate, look up info, or solve problems. And with AI Mode, you get richer answers + follow-up links.

DeepSeek open-sourced new V3.1-Terminus model
DeepSeek has released an updated version of DeepSeek v3 model, DeepSeek v3.1 Terminus. The model builds on its predecessor’s strength, showing improvements in language and agentic tool use, specifically in search and coding tasks. Terminus also shows big gains in Humanity’s Last Exam, going from 15.9 to 21.7. The model weights are open-sourced and available to download on Hugging Face.

Tools of the Trade

1. Nanobot - Open-source framework for wrapping MCP servers into AI agents that can reason, act, and maintain conversation context. It adds system prompts, tool orchestration, memory, and UI rendering (via MCP-UI) on top of plain MCP servers, letting you build richer interactive experiences.

2. Spiderseek - Track and grow your website’s visibility in AI-powered search engines (like ChatGPT, Perplexity, etc.). It provides keyword and domain research, analytics (traffic, crawls, page metrics), and content submission, so indexing happens faster. Currently, it costs $1/month in beta.

3. Claude Code Chat - A VS Code extension that brings a chat UI for Claude Code directly into the editor. It supports file referencing, context from images/screenshots, session and checkpoint history, and lets you pick the model. It also has built-in permissions and tool management so commands like shell, file ops, web fetch stay under control.

4. Awesome LLM Apps - A curated collection of LLM apps with RAG, AI Agents, multi-agent teams, MCP, voice agents, and more. The apps use models from OpenAI, Anthropic, Google, and open-source models like DeepSeek, Qwen, and Llama that you can run locally on your computer.
   (Now accepting GitHub sponsorships)

Hot Takes

1. so let me get this right:

   Oracle says Openai committed $300B for cloud compute → oracle stock jumps 36% (best day since 1992)

   Oracle runs on Nvidia GPUs → has to buy billions in chips from Nvidia

   Nvidia just announced they're investing $100B into openai

   Openai uses that money to... pay oracle... who pays Nvidia... who invests in Openai ~
   Sully
   
   

2. OpenAI runs Python in production in ways that go against all best practices and conventions you can imagine, but yeah bro Python is not production capable. ~
   Yam Peleg

That’s all for today! See you tomorrow with more such AI-filled content.

Don’t forget to share this newsletter on your social channels and tag Unwind AI to support us!

Unwind AI - X | LinkedIn | Threads

Awesome LLM Apps | Sponsor Us

PS: We curate this AI newsletter every day for FREE, your support is what keeps us going. If you find value in what you read, share it with at least one, two (or 20) of your friends 😉