- OpenTools' Newsletter
- Posts
- 🤖Alibaba's ChatGPT Clone
🤖Alibaba's ChatGPT Clone
PLUS: GPT-5.1 Released Today | World Labs Beats Google To Market
Reading time: 5 minutes
🗞️In this edition
Alibaba overhauling AI app to compete with ChatGPT
Sponsored: First AI estimator
GPT-5.1 adds personality presets and two new models
World labs just shipped its 3D world generator
Workflow Wednesday #45: Scaling Smarter
In other AI news –
Anthropic to invest $50B in US AI data centers
Parag Agrawal’s AI startup Parallel raises $100M
Russia’s AIDOL robot faceplants on debut
4 must-try AI tools
Hey there,
Alibaba's dedicating 100+ developers to rebuild its AI app because it's losing to ByteDance and Tencent.OpenAI's scrambling to fix GPT-5 with a 5.1 update just three months after launch.And Fei-Fei Li's World Labs just beat Google to market with Marble, the first commercial 3D world generator.
We're committed to keeping this the sharpest AI newsletter in your inbox. No fluff, no hype. Just the moves that'll matter when you look back six months from now.
Let's get into it.
What's happening:
Alibaba is preparing an overhaul of its main mobile AI app in coming months to help it more closely resemble OpenAI's ChatGPT, a key step in broader effort to catch rivals and eventually earn money off individual users.
Alibaba plans to start by updating existing "Tongyi" apps on iOS and Android and renaming them "Qwen," after the company's well-known AI model, people familiar with the matter said. It will then gradually add agentic-AI features to support shopping on platforms including main Taobao marketplace in coming months.
The end goal is to make Qwen a fully functioning AI agent. Alibaba plans to eventually expand globally with an overseas version. It's dedicated more than 100 developers from around the company to the makeover over past months, as part of additional AI investments that CEO Eddie Wu teased in September.
The Qwen revamp marks one of Alibaba's biggest moves to wring revenue from consumer-facing services. The Chinese e-commerce firm has joined rivals from startup Minimax to ByteDance in rolling out ever-more advanced AI models, each trying to outdo sector leaders like OpenAI and DeepSeek on performance.
The revamped Qwen app will remain free to users for now. But building a user base will help Alibaba charge for consumer-facing services in future.
In September, Wu outlined plans to roll out new models and "full-stack" AI technology, reflecting Alibaba's ambitions to both develop services and infrastructure, such as chips, that underpin technology.
Why this is important:
Alibaba trailing ByteDance's Doubao and Tencent's Yuanbao in AI app popularity is a problem for companies that dominated Chinese e-commerce.
Dedicating 100+ developers to app overhaul shows this is priority investment, not side project. That's significant internal resource commitment.
The shopping integration is Alibaba's competitive advantage. ByteDance and Tencent don't have Taobao. If Qwen becomes an AI shopping assistant integrated with China's largest e-commerce platform, that's a unique value proposition.
Keeping the app free while building a user base is classic loss-leader strategy. Monetization comes later once users are locked in. Global expansion plans suggest Alibaba sees international opportunity, not just domestic market.
Our personal take on it at OpenTools:
This is Alibaba playing catch-up while leveraging its e-commerce moat.
ByteDance's Doubao and Tencent's Yuanbao already have more users. Alibaba's late to consumer AI despite having strong AI models. The app revamp is recognition they need better consumer products, not just better models.
The shopping integration is smart play. ChatGPT can't buy things on Taobao. Qwen can. That's differentiation based on Alibaba's core asset.
But "more closely resemble ChatGPT" is admitting Alibaba's copying successful product rather than innovating. They're following OpenAI's playbook, not creating new categories.
The monetization challenge is real. "Chinese consumers are less comfortable paying for online services" is a polite way of saying free-first expectations are entrenched. How do you charge for an AI assistant when users expect everything for free?
Shopping integration solves monetization indirectly. Don't charge users for AI. Make money on transactions AI facilitates. That's a business model that works with Chinese consumer behavior.
The global expansion plans are ambitious given the app doesn't have domestic traction yet. Seems premature to plan internationally when you're losing at home.
From Our Partner:
Contractors hate estimating. It’s slow, chaotic, and half the time you’re guessing.
Estimatic AI flips that on its head.
Snap a few photos, type what needs done, and boom - you get an accurate, transparent estimate instantly based on live, local Lowe’s pricing and real labor data.
No more hunting for prices.
No more “let me get back to you.”
No more losing jobs because your competitor replied faster.
If you want to look like the smartest contractor in the room and close more bids without grinding away for hours… this is it.
👉 Try it free!
What's happening:
OpenAI is releasing GPT-5.1 today, an update to the flagship model it released in August. OpenAI calls it an "upgrade" to GPT-5 that "makes ChatGPT smarter and more enjoyable to talk to."
The new models include GPT-5.1 Instant and GPT-5.1 Thinking. The former is "warmer, more intelligent, and better at following your instructions" than its predecessor, and the latter is "now easier to understand and faster on simple tasks, and more persistent on complex ones."
Queries will, in most cases, be auto-matched to models that may best answer them. The two new models will start rolling out to ChatGPT users this week, and old GPT-5 models will be available for three months in ChatGPT's legacy models dropdown before they disappear.
As part of the update, OpenAI expanded personality presets for conversational tone. The total list now includes Default, Professional, Friendly, Candid, Quirky, Efficient, Nerdy, and Cynical. OpenAI also said it would debut an "experiment for new ways to fine-tune ChatGPT's style directly from settings," which some users will begin accessing this week.
"With more than 800 million people using ChatGPT, we're well past the point of one-size-fits-all," Fidji Simo, the company's CEO of Applications, wrote in Wednesday's Substack post.
Why this is important:
OpenAI releasing GPT-5.1 just three months after GPT-5 was released is proof enough the flagship model disappointed.
Eight personality presets from Professional to Cynical is a product feature addressing user feedback, not fundamental capability improvement. That's polish, not breakthrough.
Auto-matching queries to models suggests GPT-5.1 Instant and Thinking have different strengths. Routing complexity is OpenAI admitting no single model handles everything well.
Our personal take on it at OpenTools:
Three months from GPT-5 to GPT-5.1 is a panic update cycle.
Altman hyped GPT-5 in August. Users were unimpressed. Microsoft started using Anthropic models. Now OpenAI's shipping quick update with personality presets and routing logic.
The personality presets are lipstick on a pig. "Warmer, more intelligent, better at following instructions" is marketing speak. If GPT-5 was good enough, you wouldn't need GPT-5.1 three months later.
Microsoft using Anthropic for Copilot Researcher, GitHub Copilot, Copilot Studio, and Office Agent is an existential threat to OpenAI. That's Microsoft's $13B investment saying "your models aren't good enough for our products."
The auto-matching to models is interesting. Instead of one model that's best at everything, OpenAI's building router that picks between Instant and Thinking. That's an ensemble approach, not a single frontier model.
Keeping GPT-5 available for three months before removing it suggests compatibility concerns. Enterprise users need time to migrate. That's messier than clean upgrade.
"More than 800 million people using ChatGPT" is impressive user number but doesn't address model quality concerns. Lots of users doesn't mean users are satisfied.
The "one-size-fits-all" framing for personality presets is deflection. Users aren't complaining about personality. They're complaining about capability not meeting hype.
Releasing GPT-5.1 weeks after ChatGPT Atlas browser suggests OpenAI's throwing features at wall hoping something sticks. That's reactive product strategy, not confident roadmap execution.
Bottom line: GPT-5 disappointed badly enough to force quick update and drive largest partner to competitors. That's not winning position.
What's happening:
World Labs just shipped Marble, its first commercial world model product. Users can now turn text prompts, photos, videos, 3D layouts, or panoramas into editable 3D environments they can download and export.
The launch comes 14 months after Fei-Fei Li's startup emerged from stealth with $230 million in funding. It puts World Labs ahead of competitors like Decart, Odyssey, and Google's Genie, which are still in demo or limited research preview.
Marble differs from other world models because it creates persistent, downloadable 3D environments instead of generating worlds on the fly. Users can export as Gaussian splats, meshes, or videos. Less morphing, more consistency.
The platform includes Chisel, an experimental 3D editor that lets users block out spatial layouts (walls, boxes, planes) before AI fills in visual details. Think HTML for structure, CSS for style. You can grab objects and move them directly.
Pricing starts free (four generations) and scales to $95/month for 75 generations with full commercial rights and scene expansion features.
Why this is important:
This is infrastructure for the spatial web.
Gaming, VFX, and VR are the obvious first movers. Game developers can generate background environments and drop them into Unity or Unreal. VFX artists get frame-perfect camera control without the inconsistency plaguing AI video generators.
But the bigger play is robotics. Training data for robotics is scarce. Marble makes it easier to simulate training environments at scale. That unlocks faster iteration on robotic systems without needing massive real-world datasets.
Li's vision extends beyond gaming. She believes spatial intelligence is the missing piece for truly intelligent machines. If LLMs taught machines to read and write, world models like Marble could teach them to see and build.
Our personal take on it at OpenTools:
World Labs is moving while competitors are still demoing.
Shipping a commercial product with pricing tiers, export functionality, and editing tools means developers can actually build on this today. That matters more than flashy research previews.
The robotics angle is underrated. Simulated environments solve the data problem holding back embodied AI. If Marble becomes the standard for generating training worlds, World Labs controls critical infrastructure for the next wave of AI development.
Game developers are skeptical of generative AI, and rightfully so. But Marble's positioning as an asset generator rather than a replacement for entire pipelines threads the needle between useful and threatening.
Li's manifesto talks big about spatial intelligence changing science and medicine. That's years away. But if gaming and VFX adoption hits, the platform effects compound fast.
This Week in Workflow Wednesday #45: Scaling Smarter – Growth-Focused AI Strategies
This week, we’ll show you how to use AI Tools for clear, actionable growth insights.
Workflow #1: Auto-Summarize Growth Analytics with Claude.ai
Step 1: Export your weekly metrics from Google Analytics, Beehiiv, or Stripe as a simple .csv or screenshot.
Step 2: Drop it straight into Claude and ask:
“Analyze this data and summ……. We break down this workflow (and two more ways to scale smarter) in this week’s Workflow Wednesday.
Anthropic will invest $50 billion in building AI data centers in the US – The AI startup plans to bring data centers to Texas and New York first.
Ex-Twitter CEO Agrawal's AI search startup Parallel raises $100 million – To build web search infrastructure for artificial intelligence agents and fund deals with online content owners.
Russian Humanoid Robot Falls on Its Face in Hyped Debut – The robot, known as AIDOL, staggered onstage during a technology showcase in Moscow. Organizers blamed the mishap on calibration and lighting issues.
Viroll - An AI-powered video editing tool that helps users create highlight clips from their videos
Chekable - An AI-powered platform that provides services for checking the patentability of inventions
Book Witch - A Software as a Service platform that allows users to generate e-books from a simple description
AI Perfect Assistant - A tool designed to enhance Microsoft Word by automating various writing tasks
We're here to help you navigate AI without the hype.
What are we missing? What do you want to see more (or less) of? Hit reply and let us know. We read every message and respond to all of them.
– The OpenTools Team
How did we like this version? |
Interested in featuring your services with us? Email us at [email protected] |