DEV Community: GPTProto Official

The Top 10 Most Valuable Private Tech Companies: A Trend You Can't Ignore

GPTProto Official — Wed, 22 Apr 2026 10:37:52 +0000

As of March 2026, the list of the world's most valuable private tech companies has been released, and the figures are staggering.

First place goes to SpaceX, valued at $1.5 trillion, followed by OpenAI at $840 billion. Combined, that's nearly $2.4 trillion—more than the GDP of many countries!

But the real conversation starter isn't who ranks first or second.

Take a moment to count how many of these companies are directly related to AI. We have OpenAI, Anthropic, xAI, and Databricks—four companies right there. If you include ByteDance, which has AI deeply embedded in its operations, that number jumps to at least half.

And remember, this is just the private company list. We haven't even factored in publicly traded giants like NVIDIA, Microsoft, and Google.

Now, let’s look at another fascinating metric: revenue multiples.

Waymo boasts a staggering multiple of 315. This means the market values it at 315 times its annual revenue. xAI sits at 65.8 times, and SpaceX at 96.8 times. These companies are essentially "betting on the future." Investors aren’t just buying into current revenues; they’re buying into the vision of what’s to come.

In contrast, ByteDance has a revenue multiple of just 3.2, and Ant Group is at 3.7. While their valuations are high, these companies are genuinely making money, backed by solid profits.

To put it simply, this list showcases two types of companies: those that are valued for their imagination and those that are valued for their cash flow.

One detail that might be overlooked is Tether, coming in fourth with a valuation of $500 billion and a revenue multiple of 49. Who would have thought that a stablecoin company could secure such a high position just five years ago? The infrastructure for cryptocurrency has quietly evolved into a powerhouse.

Lastly, here’s a sobering reality: 8 out of the 10 companies are American, with only ByteDance and Ant Group representing China. And combined, their valuations still don't surpass that of OpenAI.

The gap isn't just about the number of companies; it's about the ceiling each individual company can reach.

This list is essentially telling us one thing: the greatest wealth in the next decade will likely emerge from AI and hard tech. The real question is, which side will you be on?

Don't be fooled by the allure of "sophistication"—the real pros at making presentations are leveraging AI to work smarter, not harder.

GPTProto Official — Wed, 22 Apr 2026 08:20:54 +0000

Honestly, if you're still trying to make your PowerPoint look "fancier," you might be missing the point. Last month, my team ran a little experiment: two interns prepared the same market analysis report. One toiled away for three days using traditional methods, while the other used AI tools and wrapped it up in just three hours. The result? The latter's report was praised by our boss for its clarity and vibrant data presentation—now that's what we call "sophisticated."

See, the key issue has never been about decoration.

1. My Pitfall: Treating AI like a graphic designer, and it was a disaster.

I made the mistake of thinking that "sophistication" meant finding a powerful AI tool to create a flashy cover for me. I tried it, giving the AI simple commands like, “Generate a sophisticated, techy PowerPoint.” What came out was a jumble of flashy visuals and animations that made your head spin, but the content was as empty as a company website. Show that to investors? They’d just see it as pretentious and lacking substance.

That’s the first major pitfall: if you treat AI as a designer, you’ll get designer-level results.

The real breakthrough came when I switched my approach. Instead of telling AI “how to design,” I started explaining “what I want to convey.” I gathered a bunch of meeting notes, data sheets, and even a few voice memos, then fed them to the AI.

I said, “Help me out here. This is the content for next quarter's strategy meeting. Can you identify the core issues and suggest how to present this compellingly to seasoned professionals?”

It didn’t just spit out a template; it provided me with a narrative structure: current growth is sluggish, not because of the market, but due to a critical issue in our process. Here’s the data, here’s how competitors tackled it, and here are our three solutions with their costs.

Suddenly, I had a framework. The sophistication? That’s just the muscle that grows naturally once the skeleton is in place.

2. My Time-Saving Workflow: Completing three days' work in three hours.

Forget the complicated tool names; my process is just three steps, as simple as making instant noodles.

Step 1: Argue with AI to define the core logic.

This is the only part where you need to think. Feed your ideas to AI and then challenge it. For instance, if it suggests, “We can enhance the user experience,” ask for specifics: “Are we talking about streamlining the payment process or creating a pop-up campaign? Which data shows the most significant improvement?”

Force it to squeeze real “action points” out of the “correct but vague” statements. This process transforms your fuzzy feelings into clear strategies. This strategic document is the soul of your PowerPoint.

Step 2: Let AI do the heavy lifting and generate a draft.

Take the “soul” from the previous step and simply copy and paste it into an AI PowerPoint generator. Choose the cleanest, largest-font template—don’t overthink it, just pick the simplest one.

In just two or three minutes, you’ll have a complete draft. It won’t be perfect; the layout may be basic, and the images might be average. But who cares? You now have a version you can work with instead of stressing over what image to put on the third slide of an empty document.

Step 3: Add the finishing touches where it counts.

Now you have an 80% speech draft and a 60% visual draft. Your job is to invest that remaining 20% of your effort into making it shine.

Data Slides: Replace the AI-generated basic bar charts with more impactful comparative visuals.
Core Points: Eliminate all the small text, leaving just one bold, emotional statement.
Images: If an image is crucial, spend a little time generating one that aligns with your concept. Remember, one well-placed image is enough.

3. A Simple Truth: The sophistication has always been about people, not the PowerPoint.

Tools are getting cheaper while our minds are becoming more valuable.

When your colleagues are still arguing about color schemes, you've already used AI to develop three different reporting options, testing with real data to find the most effective one. While others are burning the midnight oil polishing their presentations, you’ve practiced your speech five times and crafted engaging interactive questions using the initial draft generated by AI.

The ultimate goal of a PowerPoint isn’t the screen; it’s the nod of approval from the person across the table.

The time you save should be spent diving deeper into your business, chatting with users, and figuring out the numbers that others can't. That real-world experience, which AI can’t replicate, is the most “sophisticated” element in your presentation.

Instead of learning how to make a PowerPoint, focus on how to use AI to "outsource" the task. Reclaim your time for thinking, strategizing, and living the exciting life you deserve.

68% of AI users’ time is spent on ChatGPT—now that's a real competitive advantage!

GPTProto Official — Wed, 22 Apr 2026 04:05:21 +0000

The previous chart highlighted user engagement, showing that DeepSeek and Gemini are making significant strides.

But this chart takes a different angle—looking at total time spent—and the conclusion is strikingly different.

ChatGPT commands a whopping 68% of the total time!

When you add up the time spent across all AI products, nearly 70% of that time is dedicated to ChatGPT. Since May 2023, this share hasn’t just been maintained; it’s actually grown.

In mid-2023, ChatGPT's share was around 30-40%. Now it’s at 68%.

With more players entering the market, the leading brand’s share has surprisingly increased. This is quite rare in the world of internet products.

So, what about the other players?

Gemini comes in second with 16%, leveraging Google's ecosystem to drive traffic, so that figure isn’t shocking.

Character.AI follows with 7%, ranking third. This role-playing chat product boasts a highly engaged user base, with many young users spending hours on it daily.

The rest—DeepSeek, Claude, Perplexity, Grok—together account for less than 10%.

Wait a second, didn’t DeepSeek have the highest user engagement?

That’s right! In the previous chart, DeepSeek users averaged 25 minutes per day, surpassing ChatGPT's 18 minutes. But the issue is the user base size is vastly different.

Think of it like this: a small noodle shop has an average spend of $15 per customer, while a fast-food chain has an average spend of $5. Yet, the fast-food chain’s total revenue is several times that of the noodle shop.

DeepSeek's core users are incredibly loyal, but the total number just doesn’t compare to ChatGPT.

The harsh reality this chart reveals is that:

The AI market isn’t diversifying; it’s consolidating.

While everyone talks about a flourishing AI landscape, user time is a limited resource. When 68% of that time is concentrated on one product, it’s not just "intense competition"; it’s a clear sign of a winner-takes-all scenario.

Is Claude’s product experience good? Yes. Is DeepSeek cost-effective? Absolutely. Does Gemini have ecosystem advantages? For sure.

But user time doesn’t lie.

In the AI space, once brand recognition is established, the window for a comeback is much smaller than one might think.

AI Models API - GPT, Claude, Gemini, Midjourney etc. | GPT Proto

Unified AI API platform with GPT, Claude, Gemini, Midjourney and more. One API key for all top models. Pay-as-you-go, competitive pricing, 99.9% uptime. Built for developers and businesses.

gptproto.com

From zero to earning $40,000 a month, I hope you can learn from the pitfalls I encountered along the way.

GPTProto Official — Wed, 22 Apr 2026 02:06:19 +0000

Don’t buy into those quick-fix schemes! When it comes to launching web products internationally, there are really just three essential steps.

This time last year, I was frantically searching for projects and losing sleep over anxiety. Now? I can take a break whenever I want, while enjoying the steady stream of income notifications rolling in every day.

Honestly, I’ve tried my hand at short videos, social e-commerce, and live streaming sales. I worked tirelessly only to find that most of the profits were gobbled up by the platforms. It wasn’t until I ventured into international web products that I truly grasped the concept of passive income.

In one month, I hit income levels I had never dared to dream of. Even during holidays, my earnings continued to rise. For the first time, I experienced the surreal feeling of money still flowing in while I was taking a break.

But that’s not the main point. The key takeaway is that it took me a full year to figure out the core elements: identifying the right demand, creating the right product, and driving traffic. Let me break it down for you.

Step 1: Identify Demand

Identifying demand doesn’t mean searching for grand trends. My approach is quite simple: I browse major app charts, product communities, and plugin stores daily. I’m not just observing; I’m focusing on three things: What specific problem does this product solve? Who is using it? How does it make money?

Whenever you find something frustrating, jot it down—that’s your demand.

Determining the validity of that demand is even easier: Ask yourself, would I pay for this product? If it were created by someone I can’t stand, would I still buy it? Don’t get lost in market research; the fastest way to test your idea is to spend a few days making the simplest version and see if anyone uses it. That’s how I launched my first successful project.

Step 2: Build the Product

When it comes to building a product, forget about lofty terms like "empowerment." Focus on how to make it enjoyable for users—make it effortless for them. My principle is simple: you should be able to explain its purpose in one sentence. Hone in on one key feature and perfect it. Piling on features might impress yourself, but it won’t help anyone else.

Step 3: Drive Traffic

For free methods: Engage sincerely on forums and communities where your target users hang out, showcasing your product in various independent product spaces. For paid methods: Advertise on relevant directories, which tend to be friendly to newcomers. But honestly, don’t expect explosive traffic right away. It’s normal for a product to take months without significant buzz.

Once you reach this point, you might start seeing some income. Want to scale it? Look at a few directions: optimize your search, refine your ad targeting, partner with creators, and consistently produce content. Don’t be too shy with pricing; offer discounts for annual plans and add a couple of genuinely valuable perks.

The so-called "windfall" isn’t about chasing trends; it’s about pinpointing the right opportunity and being prepared ahead of time.

But don’t just focus on results. I’ve been through months where my income was negligible, and the pressure was real. The truth about passive income is that it’s built on countless sleepless nights in the beginning. Luck plays a role, but don’t fall for any "guru’s" foolproof formula.

Tools can enhance productivity, but don’t let them replace your own judgment as a human being.

Looking back, the income figures are just results, not the goal. The biggest gain for me has been transitioning from an "executor" to a "decision-maker." My understanding of the market, business, and myself is clearer than ever. The cycles of change are getting shorter; what works today might be outdated tomorrow. The only constant is the ability to quickly test, learn, and iterate.

So, reclaim that anxiety time and invest it back into your life with real people—like enjoying a good meal.

Thinking about investing in Anthropic? A potential 4.4x return in just four years? Here’s a peek into Coatue’s projections.

GPTProto Official — Tue, 21 Apr 2026 09:36:26 +0000

The first image shows Coatue's valuation forecast for Anthropic, while the second breaks down the investment return calculations.

In simple terms, this is a compelling case for "Why you should invest in Anthropic now."

Key Takeaways:

Entering today at a valuation of $360 billion could lead to an exit value of $2 trillion by 2030, netting a 4.4x return with an annualized IRR of 35%.
If you wait just one more year until 2031, the exit value might rise to $2.4 trillion, yielding a 5.3x return with an annualized IRR of 32%.

For an institution managing hundreds of billions, a 35% IRR is already a top-tier return. Most VC funds consider an overall IRR of 20% to be excellent.

Now, the table below is where it gets really interesting.

2026 Projections: Annual revenue of $30 billion, actual revenue of $18 billion, with an EBITDA loss of $14 billion.

2031 Projections: Annual revenue of $224 billion, actual revenue of $200 billion, with EBITDA profits of $48 billion. This implies a valuation of 41 times EBITDA.

2032 Projections: Annual revenue of $267 billion, actual revenue of $241 billion, with EBITDA profits of $78 billion. Valuation multiples drop to 31 times.

A few intriguing details:

The compound annual growth rate for revenue is projected at 70-87%. Maintaining this growth rate for five consecutive years would be unprecedented in business history. The closest example might be early AWS, but they never reached this level of growth at such a scale.
The timeline for EBITDA to turn positive is critical. Moving from a $14 billion loss to a $48 billion profit means crossing the breakeven point in a certain year, likely between 2028-2029. Until then, Anthropic will likely continue to burn cash.
Valuation multiples are expected to converge. 41 times in 2031 and 31 times in 2032. Coatue assumes that as profits grow, the market will assign more "normal" valuations. This is already a conservative stance—many AI companies today have multiples in the hundreds.

But what this presentation doesn’t address is even more thought-provoking:

How will competition play out? OpenAI, Google, and Meta are formidable players. The competitive landscape in the AI market could look very different in five years.
Can profit margins be realized? GPU costs, talent costs, data costs—any failure to control these could turn that $48 billion profit into mere speculation.
And the most fundamental question: The $360 billion entry price is built on extremely optimistic expectations.

To be frank, while the logic of this model is coherent, every single link in the chain must hold true. The premise of a 4.4x return in four years relies on everything going according to script.

But when has reality ever followed a script?

AI Models API - GPT, Claude, Gemini, Midjourney etc. | GPT Proto

Unified AI API platform with GPT, Claude, Gemini, Midjourney and more. One API key for all top models. Pay-as-you-go, competitive pricing, 99.9% uptime. Built for developers and businesses.

gptproto.com

The Truth About AI Animation: 3 Months, 4 Paths Forward

GPTProto Official — Tue, 21 Apr 2026 06:54:11 +0000

At 3 AM, I switched off the lights in my studio. Just a month before a major update from a popular AI video tool, we were still working on a lucrative project. But once the new version dropped, everything changed. Clients started slashing budgets, and the price for high-quality AI dramas was cut in half. My little studio, when factoring in labor and computing costs, suddenly had no profit margin left. It's not that we couldn't keep up; it’s just that the golden era was over.

Honestly, those posts you see about “making $10,000 a month with AI animation”? Approach them with caution—99% of them don’t tell you how cutthroat the market has become. For regular 2D projects, after paying for electricity and computing power, your earnings might be less than a typical part-time job. Unless you have steady orders that can support a team, diving in blindly now is likely just working for computing platforms and training institutions.

So, is there no way to thrive? Not at all. After reflecting, I found three viable paths to survive and even profit:

Path One: Go Global and Capitalize on Currency Differences.

Overseas platforms have a high demand for bulk live-action animations with relatively basic requirements. The key here is “bulk” and “streamlining”—turning creativity into a production line. With a solid library of prompts and efficient process management, a small team can produce substantial content. It’s all about scaling up.

Path Two: Become the “Revenue Share King” on Domestic Platforms.

Don’t rush into high-end productions; that’s a saturated market. Instead, study the revenue-sharing rules of various content platforms and create low-cost, fast-paced, and engaging “emotional quick bites.” I’ve seen small teams specialize in specific feel-good genres, driving costs down significantly. They thrive on platform revenue shares and ad placements. The essence of this industry has shifted: from a “technical job” to a “web-savvy business.”

Path Three: The Boldest—Create Your Own AI Virtual IP.

Stop being just a cog in a machine for clients. Generate a unique character using AI, continuously craft stories, and produce a series of short dramas around it. Once this IP gains traction, the premiums for commercial collaborations can be multiple times higher than standard production work. This requires not just AI skills but also an understanding of character design, content operation, and brand identity.

To be blunt, while AI has lowered the bar for creativity, it hasn’t done the same for monetization. It’s shifted the competition from “who has the best equipment or the biggest team” to “who has a stronger online presence, faster processes, and deeper platform knowledge.”

Here’s a harsh reality: many AI animation training courses teach concepts that you could learn in just a few days of engaging with high-quality content communities. They focus on basic modules but avoid discussing the real core—how to precisely control AI outputs with prompts, dissect the emotional pacing of viral shots, and significantly boost editing efficiency.

I later transitioned to co-directing at a traditional short drama company, focusing on AI-generated dramas. My biggest takeaway? Tools will always evolve. Today’s version will soon be outdated. However, the human desire for good stories and emotional connection remains unchanged.

In the world of AI animation, it's the tool users who get left behind; it’s the storytellers who thrive.

So, if you’re looking to jump in, I have one piece of advice: forget about the grand notion of “creating animations.” Start by asking yourself—can you use AI to tell a story in three minutes that a stranger would want to watch till the end, maybe even like?

If yes, then proceed. If no, save that training fee and invest your time elsewhere.

Focus on understanding human emotions instead of just tweaking parameters.

Trust me.

editor.gptproto.com

Top investment banks are predicting that Anthropic could be valued at $2 trillion by 2030. How did they arrive at this figure?

GPTProto Official — Tue, 21 Apr 2026 03:19:41 +0000

Recently, an internal presentation from Coatue was leaked.

So, who is Coatue? They’re a leading tech investment firm managing tens of billions of dollars, with investments in companies like Instacart, Snap, and DoorDash. Back in January, they created a valuation model for Anthropic.

The numbers are staggering, but when you break it down, there’s a certain logic to it.

⭕️Let’s start with the 2026 predictions:

They forecasted an annual revenue of $18 billion, with a run rate of $30 billion by year-end, but an EBITDA loss of $14 billion.

Here’s the key takeaway—Anthropic’s actual annualized revenue is already at $19 billion. Just four months into the year, they’ve surpassed Coatue's full-year estimate from January.

This means Anthropic is growing even faster than what the top investment banks predicted.

⭕️Now, let’s look at the 2031 forecast:

They project an annualized revenue of $224 billion, total revenue of $200 billion, and an EBITDA profit of $48 billion.

That’s quite a turnaround from a $14 billion loss to a $48 billion profit in just five years.

Using a 41x forward EBITDA multiple, that leads to a valuation of around $2 trillion for 2030.

🤔What does that number mean?

Currently, there are only five companies globally with a market cap exceeding $2 trillion: Nvidia, Apple, Google, Microsoft, and Amazon.

Coatue believes Anthropic could join this exclusive club by 2030.

👇However, there are a few assumptions behind this calculation:

Revenue needs to jump from $19 billion to $224 billion, a whopping 12-fold increase in five years. While it sounds ambitious, considering last year’s growth from $1.36 billion to $19 billion (a 14-fold increase), it’s not entirely out of reach—though the growth rate must remain consistent.

More critically is the shift in profit margins. From a negative 78% to a positive 24%, this indicates that economies of scale must kick in and the cost of operations needs to decrease continuously.

To be frank, the most valuable insight from this prediction isn’t the $2 trillion figure itself, but what it reveals:

Even the most seasoned analysts are being caught off guard by Anthropic’s actual growth rate.

A model created in January was surpassed by reality in April. When the actual growth outpaces predictions, it’s either a bubble or a genuine undervaluation.

Which one is it? We’ll find out by 2030.

AI Models API - GPT, Claude, Gemini, Midjourney etc. | GPT Proto

Unified AI API platform with GPT, Claude, Gemini, Midjourney and more. One API key for all top models. Pay-as-you-go, competitive pricing, 99.9% uptime. Built for developers and businesses.

gptproto.com

Don’t Let AI Do Your Thinking: The Fatal Pitfalls of Creating PPTs

GPTProto Official — Mon, 20 Apr 2026 10:34:56 +0000

Have you ever collected a bunch of “magic prompts for AI-generated PPTs”? Only to find yourself stuck when you actually try to use them? The results either look like elementary school assignments or are completely unusable.

Honestly, I’ve been there too. Last month, for an important presentation, I turned to AI for help. What happened? I spent hours tweaking it, only to end up deleting almost everything and starting from scratch. That initial draft is still sitting in the recycle bin, like a bad joke.

So, what went wrong? We missed the point entirely. AI generation is not a “magic wand.” It's more of an “amplifier.” It can’t replace your brain, but it can turn your ideas into a tangible draft at lightning speed. Expecting it to create a masterpiece with one click is like hoping a microwave can whip up a full banquet—you're using the wrong tool.

You need to be the director before AI can play the graphic designer.

So, how did I turn things around? I reviewed real projects with my team and developed a clear framework. In simple terms, it’s like having a GPS for talking to AI. For example, instead of saying, “Make a summary,” I say, “This is for the internal team; focus on how we went from A to B, including a few key challenges we faced. Make sure the content is solid, and keep it under 12 slides.”

With that clarity, the direction was right. The resulting framework scored at least a 70.

True efficiency lies within your workflow, not in some magical incantation.

My current workflow has a fixed series of steps, almost like an assembly line. First, I use smart tools as my “content assistant,” simply pouring in ideas and letting them help me structure. Next, I take the organized outline and throw it into a generation tool to create a decent draft. The third and most crucial step? I personally go through it, cutting out all the “robotic” fluff and rewriting it in my own voice. Finally, I focus on the visual polish.

This entire process can produce an 80% usable output from scratch in as little as 20 minutes. The time saved allows me to tackle more critical issues.

But here’s the catch: I almost fell into a big trap again.

Do you think that with this process, you can just sit back and relax? Wrong. The most dangerous time is when you rely too much on the tools. The examples it generates might be outdated, and the insights it summarizes could miss crucial turning points. There was one occasion where it almost presented an unverified piece of information as a conclusion, but luckily, I caught it just in time.

Tools can help you move fast, but the steering wheel and brakes must always be in your hands.

Speaking of speed, I later discovered a more fundamental issue: even with the right framework and workflow, if the “engine” you’re using isn’t powerful or comprehensive enough, or if it’s prohibitively expensive, you’ll still hit roadblocks. It’s like having the best map but driving a car with shaky performance and sky-high fuel costs—it’s futile.

We’ve since found a smarter way to handle creative and content work. Instead of getting bogged down by which single tool to use, we opted for an integrated platform like GPTProto that combines various high-quality resources. It connects top talents from different fields in a unified and straightforward way. We use one approach when we need rigorous logic and switch to another when we’re brainstorming creatively. Plus, the overall costs become more manageable through scale.

This shift has been transformative. I’m no longer “a worker at a machine,” but rather “the coordinator of a production line,” with access to the entire workshop’s resources. When tackling tasks, I can break down problems using A’s capabilities, optimize expressions with B’s strengths, and generate visual materials with C’s expertise. The whole process flows much more smoothly, and when you calculate the total costs, it’s far more reasonable than stubbornly sticking to one method.

The essence of tools is to make you the resource coordinator, not just someone executing a single function.

That’s why I say: stop getting stuck on the prompts. Prompts are the “technique,” while workflow and a holistic view are the “principle.”

What you need is not to memorize a hundred ways to press buttons, but to understand where you want to go and how to efficiently direct the entire system.

Think of tools as levers, not crutches. When you have a complete set of adaptable solutions in your hands, rather than just a single function, you’ll find yourself much more at ease when facing any challenge.

All-In-One AI APIs and AI Models for Developers | GPT Proto

Access powerful AI APIs and models with GPT Proto, including OpenAI, Claude, Gemini, and more. A fast, scalable, and cost-effective platform for developers.

gptproto.com

How much time do users spend chatting with AI every day? This chart reveals the real winners in the game.

GPTProto Official — Mon, 20 Apr 2026 09:24:03 +0000

According to data from SensorTower, six AI products were tracked for their average daily usage from May 2023 to March 2026. It's not just about who has the most users; it's about who keeps users "hooked."

This metric is a much more authentic reflection than DAU (Daily Active Users).

The standout here is DeepSeek.

By the end of 2024, it was hovering around 5 minutes, but then it took off in early 2025 and skyrocketed to 25 minutes. Right now, it boasts the highest user engagement among all AI products—hands down.

Going from 5 minutes to 25 in under a year is a fivefold increase. This growth curve is something that the others simply haven't achieved.

Gemini is another player to watch.

It started off slow, with just 3-4 minutes in mid-2024, but began gaining momentum by the end of that year and is now nearing 24 minutes of engagement—almost neck-and-neck with DeepSeek.

Don’t underestimate Google’s staying power. Once Gemini integrates into search, Gmail, and Workspace, its usage scenarios multiply, making increased user engagement just a matter of time.

What about ChatGPT?

It has grown from 6 minutes in May 2023 to around 18 minutes now, consistently increasing, but the pace is clearly not on par with DeepSeek or Gemini. In fact, the growth has flattened out in recent months.

Interestingly, as one of the earliest AI products, ChatGPT has seen its user engagement caught up by newcomers.

Claude’s situation is a bit unique.

It shot up to 16 minutes by early 2025, almost matching ChatGPT. However, its growth curve has shown some fluctuations since then.

Considering that Claude has a much smaller user base than ChatGPT, achieving similar user engagement indicates a very loyal core audience.

Perplexity remains steady.

It has consistently hovered between 5-8 minutes, with a slow but steady climb. Positioned as an AI search tool, it doesn’t require long stays—this data actually reflects a healthy usage pattern.

Grok had a rapid rise but has since declined.

It surged past 20 minutes thanks to traffic from the X platform but has recently dropped back down to around 17 minutes.

In simple terms, this chart tells us something significant:

The competition among AI products has entered a phase of "competing for user time." Whoever can get users to spend a few more minutes each day will establish a real barrier to entry.

And surprisingly, the products that are currently best at "hooking" users aren't the early starters like ChatGPT, but rather the latecomers, DeepSeek and Gemini.

The advantage of being first is losing its effectiveness in the face of user engagement.

All-In-One AI APIs and AI Models for Developers | GPT Proto

Access powerful AI APIs and models with GPT Proto, including OpenAI, Claude, Gemini, and more. A fast, scalable, and cost-effective platform for developers.

gptproto.com

Anthropic's Revenue Soars 14x in a Year: A New Contender in the AI Space Emerges

GPTProto Official — Mon, 20 Apr 2026 07:20:41 +0000

When I first saw this chart, my immediate reaction wasn’t about how strong OpenAI is, but rather—how did Anthropic pull this off?

Let’s look at the numbers.

OpenAI currently has an annual revenue of $25 billion, up from $6.25 billion just a year ago. A 4x growth is impressive, no doubt.

But what about Anthropic?

They’re now at an annual revenue of $19 billion, a massive leap from $1.36 billion last year.

That’s a staggering 14x increase.

In just one year, they’ve gone from $1.36 billion to $19 billion.

Honestly, this kind of growth is rarely seen in the history of technology.

What’s even more interesting is how quickly the gap is closing.

A year ago, OpenAI’s revenue was 4.6 times that of Anthropic. Now? It’s down to just 1.3 times.

At this rate, Anthropic could catch up to OpenAI in just a few quarters.

And let’s not forget, OpenAI had a head start of several years, with a brand recognition that’s on another level—ChatGPT is practically synonymous with AI. Despite this first-mover advantage, Anthropic is managing to close the gap. What does that tell us?

It’s all about product strength.

Anthropic’s approach is completely different from OpenAI’s. They don’t focus much on consumer marketing and instead rely heavily on enterprise clients and APIs. Claude’s reputation in the developer community has visibly risen over the past two years.

To put it simply, OpenAI had the advantage at the starting line, but Anthropic is gaining speed.

*Let’s do some quick math: *
together, they now have an annual revenue of $44 billion. Just a year ago, that number was barely $8 billion. The entire AI industry is expanding at an astonishing rate.

This isn’t a zero-sum game; the whole pie is growing rapidly.

But a 14x growth compared to a 4x growth means that if this trend continues, the landscape will inevitably shift.

Starting early doesn’t always guarantee longevity; it’s the speed that truly poses a threat.

AI Models API - GPT, Claude, Gemini, Midjourney etc. | GPT Proto

Unified AI API platform with GPT, Claude, Gemini, Midjourney and more. One API key for all top models. Pay-as-you-go, competitive pricing, 99.9% uptime. Built for developers and businesses.

gptproto.com

From Monthly Updates to Daily Posts: Building an AI Knowledge Base

GPTProto Official — Fri, 17 Apr 2026 10:56:44 +0000

At 3 AM, I found myself still writing a "resume" for my AI. During that time, I felt a bit ridiculous. I kept copying and pasting my WeChat official account positioning, writing style, and product structure into different AI chat boxes. Each time, I had to start with, "Hello, I'm Xiao G, and my style is..." Then I'd close the chat and start over.

A month later, it hit me:** I wasn't using AI; I was just being its assistant.**
So, I spent some time testing mainstream AI knowledge base tools on the market—like Feishu, Notion, IMA, and NotebookLM. The conclusion? Each has its pros and cons, but most suffer from two major issues: either the AI can't access all of my notes, or it can't directly manipulate my local files.

In the end, I opted for the Obsidian + Claude Code combination. The reason was simple: it allows AI to truly integrate into my workflow. I gathered all my materials—hundreds of drafts, recordings, and product documents—into one place. The AI can see everything and can directly modify the content.

It's worth noting that the AI models powering this system aren't fixed. I use Claude for writing, switch to Midjourney for images, and might employ Kimi for processing long audio. If I had to apply for separate APIs for each model and handle different calling formats, efficiency would plummet.

That's where aggregation platforms like GPTProto come in. To put it simply, they consolidate multiple large model APIs into a single entry point. The workflow I built in Claude Code can flexibly call the most suitable model behind the scenes without major overhauls. This keeps costs manageable and stability assured.

The results were immediate.

The most noticeable change was with my WeChat account. Previously, writing a few thousand words would take an entire day. Now, from topic selection to final draft, it takes about an hour. The quality hasn't diminished; in fact, the metrics have improved. The biggest shift from monthly to daily updates isn’t about working harder; it’s that the tools have made daily updates feasible.

I've also streamlined my Xiaohongshu posts, consistently producing around 20 notes a day. Some may not get high views, but they still convert. The AI has learned my tone and style, creating content that resonates with readers.

What truly made me feel like I had "arrived" were two key achievements.

**First, I created two virtual products for sale—one priced at 39 and the other at 599. **From concept to copy and detail pages, all of it was completed through the collaboration of AI modules in my knowledge base. It acts not just as a writer, but also takes on some of the responsibilities of a product manager.

Second, I attended an in-person class and fed several hours of recordings into the system. By the end of the afternoon, it had helped me organize over a dozen core methodology documents. When I write industry analyses, I can directly cite specific insights from the instructors. Knowledge transformed from a static archive into a resource readily available for use.

That’s the key. Many people use AI as a search engine, asking one question at a time. But my approach turns AI into a "digital partner." I predefine collaboration principles, personal guidelines, and target directions in the configuration file. In the workflow folder, I outline the execution steps for various tasks, like the complete process for "writing a Xiaohongshu note."

It doesn’t need to ask, "Who am I?" every time. It knows.

I also implemented an "iteration mechanism": have the AI generate two versions of the copy—one for me to tweak manually, and the other to remain unchanged. Then, I provide feedback by comparing the modified version with the original. After a few rounds, the content it generates gets closer to my desired style. This isn’t just about fine-tuning a tool; it’s about training an assistant that understands me better.

Of course, this method has its challenges. It requires a certain level of file management skills and some time to set up the initial framework. The early stages are an investment, not a quick-fix solution.

But once it’s up and running, the results are sustainable.

Now, I spend less than a third of the time I used to on content creation. I use the saved time for thinking, collaborating, and developing new products. A friend’s ten-person team generates less monthly revenue than I do alone. It’s not because I’m exceptionally talented; it’s because I found the right leverage.

The purpose of tools isn’t to make us busier; it’s to give us more freedom.

That person who was up at 3 AM writing a "resume" for AI can finally shut down the computer and get some sleep.

All-In-One AI APIs and AI Models for Developers | GPT Proto

Access powerful AI APIs and models with GPT Proto, including OpenAI, Claude, Gemini, and more. A fast, scalable, and cost-effective platform for developers.

gptproto.com

Stop struggling with teaching materials! Teachers are making six figures a month from AI-powered public classes. Can you believe it?

GPTProto Official — Fri, 17 Apr 2026 07:14:53 +0000

Honestly, I didn’t at first either. Before 2025, I was selling teaching materials on Xiaohongshu, working tirelessly to attract hundreds of visitors daily, with a conversion rate barely scraping 10%. That all changed when I discovered the “AI Empowered Public Classes.”

This isn’t just a trend; it’s a necessity. Teachers need public classes for promotions, competitions, and school inspections. Preparing a single lesson can take two to three days, and it can be exhausting. What I’ve done is compress that work into just three hours.

✅Step one: Break down the script and create storyboards.

Don’t overthink it; just use AI to generate the script for the “scenario introduction” part of the public class and then turn it into a video storyboard. For example, I broke down the elementary school text “Little Tadpoles Looking for Mom” into five storyboard frames, making it ready for teachers to use immediately.

✅Step two: Revamp the PPT and adjust the lesson plan.

There are countless public class PPTs out there, but they all look the same. I change the color schemes, tweak animations, and add interactive designs. The same content can have three times the visual impact with these adjustments.

And that’s it! The results? Out of every ten teachers who contacted me for consultations, four went ahead and made payments. The unit price isn’t high, but the volume is substantial. With my Xiaohongshu store and private consultations, I’ve consistently maintained six-figure monthly sales.

But there’s a catch. 👇
You think AI can do everything automatically? Dream on. The scripts generated by AI need editing, storyboards require adjustments, and PPTs need to be creatively revamped. If you’re hoping to just sit back and let this run on autopilot, it’s time to wake up.

This market had only 50-100 competitors last year, but it’s doubled this year. Still, compared to the saturated teaching materials market, this space is relatively untapped.

What strikes me the most isn’t just the money. 🤔
One teacher reached out to me privately, saying she won the city’s first prize using my public class and got promoted to a senior title. She said, “Your materials have made a huge difference in my career development.”

In that moment, it hit me: We’re not just selling materials; we’re selling time, opportunities, and a ladder for ordinary people to climb higher.

Teaching materials help students improve their scores, while public classes help teachers advance their careers. The latter is a more pressing need, more sustainable, and definitely worth pursuing.

Tools may become outdated, and AI will continue to evolve.

But the desire for self-improvement in people will always remain.✨

Let’s give back the time teachers spend preparing lessons, especially those who are still up late revising their plans.