The Revelation of Mythos Closure: Renting AI or Owning It Yourself — Which Is Better? Control Determines Life or Death

Mythos Shutdown Reveals Core Issues in AI Entrepreneurship: When Products Are Built on External Models, What Do Companies Truly Own?
(Background: Security Expert: AI models as dangerous as Claude Mythos are unstoppable, will be everywhere within 24 months)
(Additional context: Anthropic's strongest AI restricted from export by the US! Fable 5, Mythos 5 go offline globally)

Table of Contents

Toggle

• Mythos Shutdown: Hidden Risks of Renting Intelligence
• Renting vs. Owning: Risks of API Dependency
• Definition of True Ownership: Data and Workflow Accumulation
• The Era of Multiple Frontiers: No Single Model Monopoly
• Product Launches: K2.7 Code, M3 Simultaneously Released

This article is based on analysis by user lqiao on x.com.

The shutdown of Mythos this week has caused many AI entrepreneurs to revisit a core issue obscured by cost considerations. When product capabilities are built on external models and platforms, what does the company truly control? In recent years, open-source models have often been viewed as cheap alternatives, but this article argues that control is the key variable. Calling APIs can quickly launch products, but it also means that core capabilities are constrained by vendor rules. The real moat is not about calling more powerful models, but about transforming intelligence into proprietary assets.

In recent years, open-source models have frequently been discussed within the framework of "cheaper frontier model alternatives." But this article suggests that cost is not the most critical factor; control is. For an AI company, calling frontier model APIs can rapidly enable product deployment and lower technical barriers, but it also means that core capabilities may be subject to the model provider’s rules, pricing, strategic adjustments, or even takedowns.

The article further emphasizes that "owning intelligence" does not mean abandoning frontier models, but rather that companies should embed their data, workflows, domain knowledge, evaluation standards, and edge cases into a controllable model system. The future of AI competition will not necessarily be dominated by a single largest model, but by multiple "frontiers": general-purpose frontier models, proprietary fine-tuned enterprise models, vertical-specific models, and routing systems that coordinate multiple models.

Mythos’s shutdown serves as a reminder: the true moat in the AI era is not just about calling more powerful models, but about turning intelligence into the company’s own asset.

Mythos was shut down this week. Whether you agree with this decision or not, that’s no longer the point.

What truly stings many is: a company built on intelligence it cannot control suddenly finds itself under a set of decisions it cannot influence. After seeing this, many founders ask themselves the same question: which parts of my business are actually just "rented"?

Mythos Shutdown: Hidden Risks of Renting Intelligence

Over the past few years, discussions about open-source models have mostly focused on costs: can they really accomplish tasks? If so, how much cheaper are they compared to calling frontier model APIs?

By now, we have a fairly clear answer. We have collaborated with companies like @RampLabs, @cursor_ai, @harvey, and others, following similar approaches: starting from a powerful open-source model, fine-tuning it on the most critical parts of the business, and continuously evaluating it against frontier models.

The results have repeatedly surprised us. For the most important tasks for enterprises, a fine-tuned open-source model can often approach or even match the quality of frontier models at very low cost.

But what truly became clear this week is: cost has never been the most important issue.

The deeper issue is control. Who owns the intelligence your product depends on?

Recently, many discussions have been summarized as the difference between "renting" and "owning." The analogy isn’t perfect, but it’s very useful.

Renting has always worked well before problems arise. Apartments are move-in ready, lights work, water runs, repairs are handled. That’s why most companies choose this path initially.

Renting vs. Owning: Risks of API Dependency

Frontier model APIs are excellent products. They enable startups to build things that seemed impossible just a few years ago.

But renting also means limitations. Landlords can raise rent, decide what modifications you can make, change the rules. Occasionally, for reasons unrelated to you, they can even tell you: you have to move out.

You haven’t done anything wrong. You’re just operating on someone else’s turf.

That’s why Mythos’s story resonates with so many. When your core capabilities depend entirely on someone else’s platform, you’re exposed to decisions beyond your control.

Most of the time, this isn’t critical. But sometimes, it becomes extremely important in an instant.

The lesson isn’t that companies should stop using frontier models. Far from it. Frontier model labs have achieved extraordinary technology. Most products should use them. We ourselves are using them.

In many ways, frontier models are becoming infrastructure. But infrastructure and ownership are two different things.

Definition of True Ownership: Data and Workflow Accumulation

You can use public infrastructure while still owning what truly creates value for your business. In AI, "ownership" means starting from a cutting-edge open-source model and shaping it around your company’s most unique aspects.

Your data.

Your workflows.

Your domain knowledge.

Your edge cases.

Your evaluation standards.

Your definition of "good."

The Era of Multiple Frontiers: No Single Model Monopoly

Over time, this model will become less general-purpose and more capable of reflecting the actual work your company handles daily. That’s where value is created.

Think of it as a house. Moving furniture is easy, repainting a wall is easy. But if your future depends on the layout of the house itself, you’ll eventually want the ability to move walls. The same applies to intelligence.

When intelligence truly belongs to you, no one can quietly remove the floor beneath your product.

That’s why we build Fireworks this way.

We place training and inference within the same system, allowing companies to adopt the best open-source models, shape them around their most critical issues, and deploy steadily into production.

Not just consuming intelligence. Owning intelligence.

There’s also an optimistic lesson this week: the future of AI isn’t about one model winning everything.

Product Launches: K2.7 Code, M3 Simultaneously Released

There’s no such thing as a single frontier. There are many frontiers.

A frontier model is one kind.

A model fine-tuned with years of proprietary company knowledge is another.

A specialized model that solves a narrow problem better than any other is another.

A system that routes requests to multiple models, enabling collaboration and surpassing single models on many tasks, is also a frontier.

The most exciting change in AI isn’t that one model is becoming smarter, but that intelligence is becoming more customizable.

The ultimate winners won’t necessarily be those with the largest models, but those who can turn intelligence into their own unique assets.

Much of this week has been spent reacting to news, but we continue to release products: @Kimi_Moonshot K2.7 Code, @MiniMax_AI M3, @Alibaba_Qwen 3.7 Plus.

My hope for the future isn’t a single model quietly consuming everything it sees.

But many teams owning their own part of the frontier.

If Mythos’s shutdown prompts you to rethink the trade-offs involved, we’re happy to chat.

[Original link]