The Next Wave – 5 New AI Models That Are Smarter (and Cheaper) Than GPT-5

OpenClaw and the AI Agent Revolution

OpenClaw—originally ClaudeBot—is an AI agent that acts almost like a digital employee. Peter Steinberger created it and got so much attention he ended up at OpenAI, after a wild name-change drama with Anthropic. Steinberger’s tools made it possible for anyone to set up an agent that connects with messaging apps like Telegram, Slack, Discord, and WhatsApp, handling tasks ranging from emails to scheduling to automating workflows.

OpenClaw lets users grant different levels of access, so your AI agent could check your calendar, read emails, or even control apps on your behalf. Steinberger decided not to build a billion-dollar company by himself. Instead, he picked OpenAI for its promise to keep OpenClaw open-source and let people use whatever AI models they prefer.

Key Takeaways:

• Agents are not just browser gimmicks, but real digital helpers.
• OpenClaw is open-source, now outside Steinberger’s direct oversight.
• Anthropic fumbled by blocking Claude Max plan users and lost momentum.
• OpenAI’s welcoming attitude fuels agent adoption and community support.

Meta and Manus: Agents in Messaging & Ads

Meta jumped into the agent game with Manus AI, letting users integrate agents in messaging apps, starting with Telegram and soon WhatsApp. Manus agents focus on automation, especially in ads. Meta also announced Manus for Ads Manager, making ad creation and workflow management simpler for business owners.

Agents create ad creatives, adjust budgets, and optimize for conversions, reducing manual labor. While some tasks are just logic statements (“If cost per lead > $2, turn ad off”), Manus takes the friction out by automatically handling creative production.

Key Points:

• Manus AI supports Telegram first, with plans for WhatsApp.
• Business owners can prompt Manus to create ads, analyze performance, and adjust campaigns.
• Agents are becoming the “feature” everyone wants, even when not strictly necessary.

Agent Costs vs. Employee Costs

Mark Cuban challenged the hype, asking: What happens when AI token costs exceed what you’d pay an employee? In coding and automation, AI token spending can add up. However, the cost is dropping quickly because new models are dramatically cheaper and just as capable as older, expensive ones.

Key Stats:

• The state-of-the-art coding models are becoming accessible at fractions of the previous price.
• Sonnet 4.6 does Opus-level work at one-third the cost, making the price gap shrink fast.
• Big token spending is a short-term issue; new models will disrupt it soon.

New AI Models: Smarter and Cheaper Than Ever

This month has been the single biggest period for AI model releases in history. At least five new models hit the market, including Claude Sonnet 4.6, Gemini 3.1 Pro, Grok 4.2, ByteDance’s Seed 2.0, and Qin (from China). These models deliver higher quality tasks, bigger context windows, and lower prices.

Benchmarks:

• Sonnet 4.6 handles 1 million tokens, letting users feed it tons of data.
• Gemini 3.1 Pro leads in science, logic reasoning, and competitive coding.
• Grok 4.2 stands out with its “council of experts,” combining four models for a consensus answer.

Casual users benefit as top-tier intelligence enters the free or affordable tiers. File creation, connectors, skills, and more are now bundled at no extra cost.

Key Takeaways:

• AI models are getting smarter, but also much cheaper and more accessible.
• Context windows mean you can process entire books, big datasets, and huge transcripts.
• The mainstream consumer benefits from “Opus-level” performance in affordable models.

Creativity, Human Taste, and AI Content

AI can make videos, music, graphics, and even write entire scripts, but most people still make basic stuff—or “slop,” as Matt Wolfe called it. Only those with taste, opinions, or unique vision will truly stand out.

Fei-Fei Li, aka “the godmother of AI,” says AI should superpower human creators, not replace them. Real creativity comes from the human stories, values, and ideas we bring. Hollywood and the music industry are debating whether AI can replace creative jobs, but voices like Matthew McConaughey and Ice-T advise creators: own your IP, set your boundaries, and use AI for productivity.

Examples:

• Ice-T uses AI for music videos, saving money and time.
• McConaughey trademarked his voice and likeness, letting studios pay to use it.
• AI only augments creativity—if you bring the vision, AI can handle production.

Business Tools: Visual Branding and Dropshipping

AI tools for branding are changing how businesses create and iterate. Google’s Pomeli Photoshoot lets you upload a product photo, builds out full brand kits from URLs, and automatically generates photo shoots with your product in different settings. You can even make animated videos with recognizable backgrounds (like Petco Park for the Padres).

You can now launch a dropshipping business in hours, grabbing photos from AliExpress, using Pomeli for product imagery, setting up Shopify, and automating ads and customer service with AI. The barriers for entry are almost gone.

Key Takeaways:

• Pomeli speeds up product photography, branding, and campaign creation for e-commerce.
• Dropshipping is easier than ever thanks to AI-generated assets and marketing tools.
• Business DNA feature acts as custom instructions for branding and content.

Robots, Manufacturing, and the Balance of Power

China’s Spring Festival Gala featured robot ninjas doing martial arts, showing just how far mechanical engineering has come. The hardware is there—robots can do backflips, handle weapons, and synchronize movements. The challenge is software—robots still struggle to understand their environment, adjust force, and handle household tasks.

Elon Musk’s Optimus robot relies heavily on parts from China, highlighting the difference in manufacturing culture. The U.S. leads in AI compute; China leads in cheap, fast robot production. If those strengths combined, the world would see super-smart, mass-produced robots.

Key Points:

• Chinese robots outperform in mechanical feats, but lag in intelligence.
• U.S. holds the chip manufacturing and AI design advantage.
• Speed of manufacturing in China allows for rapid iteration and improvement.

Key Links and Resources

• The Next Wave Podcast: https://www.thenextwave.show
• Matt Wolfe: https://www.youtube.com/@mreflow 
• Seedance 2.0: https://www.seedance.com/
• OpenClaw: https://openclaw.ai/
• Manus: https://manus.im/
• Nano Banana: https://nanobanana.com/
• ElevenLabs: https://elevenlabs.io/
• Perplexity: https://www.perplexity.ai/

Conclusion and Top Highlights

AI is moving faster than ever, and this episode showed how agents, models, branding tools, and robots are making things smarter and cheaper. OpenClaw’s shift to OpenAI marks a new era for agents you can trust. Business owners can launch whole stores and ad campaigns overnight. Creators can use AI as a super-tool—not as a replacement. Models are leveling up and prices are dropping, helping everyone from tech geeks to total beginners.

Agents are finally useful, the model race is far from over, and creativity and taste still matter more than ever. Stay up-to-date, keep leveling up your skills, and use these tools to work smarter—not just harder!