Well…this week proved that even in AI's wild west, not every launch is a home run. While China dropped a GPT-5 contender and ElevenLabs went full music mogul, OpenAI learned the hard way that 700 million users have opinions: strong ones.
Quick Wins (5 Minutes Each)
DeepSeek V3.1 — Try: Visit deepseek.com and test the Chinese GPT-5 rival that's optimized for different hardware, costs less, and might just be better than OpenAI's latest.
ElevenLabs Music Generator — Create: "Dreamy indie rock with reverb-soaked vocals" → Generate royalty-free commercial music that actually sounds professional.
GPT-5 Reality Check — Experience: See why Sam Altman admitted they "totally screwed up" and had to bring back GPT-4o after user revolt. Was there a screw-up, or are users just prompting wrong?
Pick one, spend five minutes, and witness AI's most dramatic week of wins, losses, and honest confessions.
DeepSeek V3.1 — The Free AI That's Competing GPT-5
What Just Happened: China's DeepSeek just dropped V3.1: completely FREE to use and matching (sometimes beating) GPT-5's performance on key benchmarks.
The Money Game: While OpenAI charges monthly for ChatGPT Plus, DeepSeek V3.1 is:
100% free for basic use
10x cheaper than GPT-5 for API calls
No usage limits for most tasks
Open source so you can run it yourself
What's New from DeepSeek V2:
3x larger at 685 billion parameters (V2 was 236B)
Better reasoning - now combines fast answers + deep thinking in one model
Longer memory - 128K context (can process entire books)
Smarter efficiency - only activates the parts it needs, keeping costs low
Real-World Performance: DeepSeek V3.1 actually outperforms GPT-5 on coding tasks and matches it on most business applications—while being completely free. However, the DeepSeek app lacks much of the extra features many love on ChatGPT.
Test out DeepSeek’s new power with some of our advanced prompts from the
Business Impact: Imagine cutting your AI costs by 90% while getting better results. That's what DeepSeek V3.1 offers. However, do your due diligence about the security with this AI model before implementing into your business..
The GPT-5 Fiasco : Why Great Prompting Still Wins
The Confession: Sam Altman (CEO of ChatGPT) admitted: "I think we totally screwed up some things on the rollout."
What Went Wrong: GPT-5's colder, more "robotic" personality triggered an unprecedented user backlash. But here's what most people missed—the model is actually MORE powerful, it just needed better prompting.
The Hidden Truth: While users complained about GPT-5 feeling "corporate," advanced users discovered that with the right prompts, GPT-5 could be warmer, more creative, and more helpful than GPT-4o ever was.
A Business Lesson: This proves why excellent prompting skills are now your competitive advantage. While beginners struggle with basic interactions, skilled prompters are getting incredible results from the same model.
What Advanced Users Know:
GPT-5 responds better to specific personality instructions
Adding context like "respond conversationally and warmly" transforms the output
System prompts can make GPT-5 behave exactly how you want
The model's raw capabilities are actually superior—it just needs guidance
The Prompt Engineering Opportunity: While others complain, smart business owners are mastering prompts to get better results than ever before. GPT-5's power is still there: you just need to know how to unlock it.
The Bigger Reveal: Altman admitted OpenAI has even better models but can't release them due to capacity limits: "We have better models, and we just can't offer them, because we don't have the capacity."
Try This Week: Instead of switching back to GPT-4o, experiment with GPT-5 optimized prompts. Wrote an amazing guide on prompting with GPT-5:
Must Read: Make sure to check out ’s recent article on GPT-5 & her excellent prompt guide for GPT-5:
ElevenLabs Music — Your New Revenue Stream
The Business Opportunity: ElevenLabs just launched commercial-ready AI music generation that could save your business thousands in licensing fees while opening new revenue streams.
The Smart Legal Strategy: Unlike competitors getting sued, ElevenLabs struck licensing deals with major publishers representing Adele, Nirvana, Beck, and Childish Gambino. This means you can actually use the music commercially without legal worries.
Real Applications:
Content Creators: Generate custom background music instead of paying $50-200 per track
Marketing Agencies: Create brand-specific jingles and soundtracks for clients
Video Producers: No more copyright strikes or expensive royalty-free music subscriptions
App Developers: Custom music for games, apps, and experiences
Course Creators: Unique intro/outro music that matches your brand
Revenue Opportunities: Smart entrepreneurs are already:
Offering custom music creation services to local businesses
Building music libraries for specific niches (fitness, meditation, corporate)
Creating subscription-based music services for small businesses
Developing AI music apps using ElevenLabs' API
Quality Examples:
"Dreamy, psychedelic, slow Indie Rock, reverb-soaked vocals, retro keys"
"Epic cowboy show track, wild west, cinematic orchestral elements"
"Upbeat corporate background music, motivational, clean production"
Try This Week: Calculate how much you spend annually on music licensing. Then generate 10 custom tracks for your business and compare the quality to what you're paying for.
Manus AI — The Autonomous Agent That Actually Works
The Hype Train: Everyone in AI is talking about Manus. We put it to the test. The new general AI agent from China had some system crashes and server overload—but it's highly intuitive and shows real promise for the future of AI helpers.
What Makes It Different: Manus is a fully autonomous AI system designed to run asynchronously in the cloud—no repeated prompts, no babysitting.
The Latest Updates (August 2025): Explore the latest Manus AI 2025 update featuring GPT-5 integration, advanced image editing, and 24/7 autonomous task execution designed to save businesses time and boost growth.
Real-World Capabilities: By combining parallel processing, agent-to-agent collaboration, and real-time updates, it simplifies complex tasks and enables professionals to achieve their goals with greater efficiency. As mentioned earlier, try our & take our course to get the most out of Manus AI.
Try This Week: Sign up for the beta and test autonomous task execution—let it handle research, analysis, and workflow automation while you focus on strategy.
This Week's Reality Checks
1. The Personality Problem is Real: GPT-5's technical improvements meant nothing when users felt emotionally abandoned.
2. China's AI Strategy is Working: DeepSeek V3.1 proves you can build world-class AI without Silicon Valley or Nvidia.
3. Legal Strategy Matters: ElevenLabs' licensing approach might be the template for AI companies entering creative industries.
4. Infrastructure > Hype: While everyone debates AI model capabilities, the real money is in making AI actually work for businesses.
5. We're in an AI Bubble: Even Sam Altman (CEO of OpenAI) admits investors are "overexcited about AI"—but also that "AI is the most important thing to happen in a very long time."
💭 The Week's Uncomfortable Questions
Are we too emotionally attached to AI? The GPT-5 backlash suggests people prefer their AI companions to have personalities.
Is China winning the AI race? DeepSeek V3.1 matches GPT-5 at a fraction of the cost, optimized for their own chips.
Will AI ruin human creativity? ElevenLabs can now generate music that sounds indistinguishable from human artists.
Are we building the right infrastructure? Altman says OpenAI needs to spend "trillions" on data centers: is that sustainable?
✅ This Week's Action Items
Test DeepSeek V3.1 — Compare it to GPT-5 on your hardest problems
Experiment with ElevenLabs Music — Generate royalty-free tracks for your projects
Audit Your AI Stack — Are you using the right models for each task?
The Bottom Line
This week proved that AI success isn't just about bigger models or better benchmarks—it's about understanding humans. While China builds strategic alternatives and companies navigate legal minefields, the biggest lesson came from 700 million ChatGPT users who reminded Silicon Valley that relationships matter more than algorithms.
The Trend to Watch: We're transitioning from the "AI can do anything" phase to the "AI needs to do it right" phase. Technical capabilities are table stakes—user experience, legal strategy, and emotional intelligence are the new differentiators.
This week reminded us that in AI, as in life, it's not just what you build—it's how people feel about what you build. Sometimes the most important metric isn't on any benchmark.
