Original Title: Owning vs. Renting Intelligence
Original Author: Lin Qiao
Translation: Peggy, BlockBeats

Editor's Note: Mythos was shut down this week, reminding many AI entrepreneurs of a question obscured by cost discussions: When a product's core capability is built on external models and platforms, what does the company truly own?

In recent years, open-source models have often been discussed within the framework of "cheaper alternative frontier models." But this article argues that cost is not the most critical variable; control is. For an AI company, calling API access to cutting-edge models can quickly launch products and lower technical barriers, but it also means that core capabilities may be subject to the rules, prices, strategic adjustments, or even takedown decisions of model providers.

The article further suggests that "owning intelligence" does not mean abandoning frontier models, but rather that companies should embed their own data, workflows, domain knowledge, evaluation standards, and edge cases into a controllable model system. The future AI competition may not be dominated by a single largest model, but will feature multiple "frontiers": general frontier models, enterprise-specific post-training models, vertical niche models, and routing systems composed of multiple models working together.

The shutdown of Mythos serves as a reminder: the true moat in the AI era is not just how powerful a model you can call upon, but whether you can turn intelligence into your company's own asset.

Below is the original text:

Mythos was shut down this week. Whether you agree with this decision or not, that’s not the point anymore.

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

In recent years, discussions about open-source models have mostly focused on cost: can they really perform tasks? If so, how much cheaper are they compared to calling frontier model APIs?

Now, we have a fairly clear answer. We have collaborated with companies like @RampLabs, @cursor_ai, @harvey, and the general approach is similar: start from a powerful open-source model, fine-tune it on the tasks that truly matter to the company, and continuously compare it with frontier models through strict evaluation.

The results are consistently surprising. For critical business tasks, a fine-tuned open-source model can often approach or even match the quality of frontier models at a very low cost.

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

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

Recent discussions are often summarized as the difference between "renting" and "owning." The analogy isn't perfect, but it’s very useful.

Renting Intelligence

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

API access to frontier models is an excellent product. It allows 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 also tell you: it’s time 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 expose yourself to a set of decisions beyond your control.

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

Owning Intelligence

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 are also using them ourselves.

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

You can use public infrastructure while still owning what truly creates value for your business. In AI, "owning" means starting from an advanced open-source model and shaping it around your company's most unique parts.

Your data.

Your workflows.

Your domain knowledge.

Your edge cases.

Your evaluation standards.

Your definition of "good."

Over time, this model will become less general and more reflective of the actual work your company handles daily. The value is created right here.

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 silently remove the floor beneath your product.

That’s also why we built Fireworks this way.

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

Not just consuming intelligence. But owning intelligence.

There is no single frontier

Another optimistic insight this week: the future of AI does not depend on one model winning everything.

There is no single frontier. There are many kinds of frontiers.

A frontier model is one kind of frontier.

A model fine-tuned on years of proprietary company knowledge is another kind of frontier.

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

A system that routes requests to multiple models, allowing them to collaborate and outperform single models on many tasks, is also a frontier.

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

The companies that ultimately succeed won’t necessarily own the largest models, but those that can turn intelligence into their own unique assets.

Looking Ahead

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

My ideal future isn’t one where a single model quietly consumes everything it sees.

It’s one where many teams can own their own part of the frontier.

If Mythos’s shutdown has made you reconsider the trade-offs involved, we’re happy to chat.

[Original Link]

Click to learn about Rhythm BlockBeats job openings

Join the Rhythm BlockBeats official community:
Telegram Subscription Group: https://t.me/theblockbeats
Telegram Discussion Group: https://t.me/BlockBeats_App
Twitter Official Account: https://twitter.com/BlockBeatsAsia