First Impressions After 24 Hours with Fable: The Token-Rich World and What It Means for Everyone

After a full day of intensive use, Fable is not the same model we had in June. It’s evolved — sharper, more reliable, and still the strongest frontier model available today. But the specific improvements to the model itself are almost secondary now.

What matters far more is how dramatically the world has already split into two groups: those who can afford to spend heavily on tokens and those who cannot.

First Impressions After 24 Hours with Fable: The Token-Rich World and What It Means for Everyone In just two days of heavy usage, we burned through several thousand dollars. Roughly 90% of that spend went toward saving cognitive load and reducing friction — Fable completes tasks more accurately, requires far less back-and-forth clarification, and simply gets things right on the first or second try.

The remaining 10% unlocked genuinely new capabilities — things that were effectively impossible with even the best previous models (including Claude Opus).

This cost structure already reveals deeper second-order effects that every founder, executive, and knowledge worker should be thinking about.

1. Digital Products Are Getting Much Harder to Build Profitably

We’ve become deeply skeptical of most ideas for new purely digital products. Digital businesses aren’t going to disappear overnight, but the competitive bar has risen dramatically. Without some kind of unfair advantage — proprietary data, distribution, regulatory moat, or extreme execution speed — standing out will be extremely difficult.

First Impressions After 24 Hours with Fable: The Token-Rich World and What It Means for Everyone The realistic paths forward appear to be:

Building things tightly connected to the physical/offline world, or
Backing (or joining) teams that are genuinely committed to dominating their specific niche with AI-native approaches.

Everything else risks becoming commoditized very quickly.

2. AI Access Is No Longer Equal — And This Is a Social, Not Just Technical, Shift

First Impressions After 24 Hours with Fable: The Token-Rich World and What It Means for Everyone The best models will increasingly be reserved for the Pentagon and its Chinese counterpart. One tier down will be large corporations willing to spend $100 million+ per year on tokens. What remains will trickle down to everyone else.

This isn’t primarily a technical limitation anymore — it’s an economic and strategic one. If your competitor has access to a significantly more capable or “safer” model, that creates a structural disadvantage that’s hard to overcome with traditional business tactics. The gap isn’t just about who has the latest model; it’s about who can *sustain* access to frontier-level intelligence at scale.

3. Agents Have Crossed a Real Autonomy Threshold

Fable’s agents are now genuinely capable of long, independent work sessions. Our average productive session exceeds 80 minutes.

Within that time, the model will:

Propose and run multiple experiments;
Critically evaluate the results;
Iterate on its own;
Spin up sub-agents for research when needed;
Recover from dead ends without constant human intervention.

This is a qualitative leap, not just an incremental improvement. We’re moving from “very helpful assistant” to “capable junior colleague who can drive projects forward for extended periods.”

4. In Six Months, Everyone Will Have Access to Today’s Frontier

Here’s the uncomfortable part: within roughly six months, we expect a powerful open-source (or cheaply accessible) model at roughly the level of current Fable to be widely available.

The absolute frontier will have moved further ahead by then, but the floor will have risen dramatically. This means that not only Anthropic’s top enterprise clients, but also a spam operation in Pakistan, a small marketing agency, or an individual freelancer will have access to intelligence that was previously reserved for well-funded organizations.

What does that mean for your industry? Your customers? Your own job or business model?

The democratization of high-level capability is coming faster than most people expect — and it will be messy.

5. Most Companies Are Still Wasting Enormous Amounts of Human Time

Despite years of warnings, the majority of organizations still haven’t internalized what’s happening. Across nearly every industry we look at, we see hundreds of thousands of human hours being spent on work that simply doesn’t need to exist in an AI-native world.

This isn’t low-quality work — quite the opposite. It’s often high-quality, diligent effort being applied to tasks that AI can now handle faster, more consistently, and at lower marginal cost.

Most companies have not yet restructured their workflows, roles, or expectations around this new reality. That lag represents a real, time-limited opportunity for those willing to move aggressively. It won’t last forever.

---

Bottom line: Fable (and models like it) aren’t just better tools. They’re forcing a rapid re-evaluation of what’s worth building, who has power, how work gets done, and how quickly advantages erode.

The token economy is here.

The question is no longer whether you’ll use these systems — it’s how quickly you can restructure around them before your competitors do.

The next six to twelve months will separate the organizations that treat frontier AI as a core strategic asset from those that treat it as just another productivity tool. The gap between those two groups is already widening fast.

---