Back to Devoured - August 24, 2026
DEVOURED
Fable & The End of the Free Lunch

Fable & The End of the Free Lunch

Tech aillminfrastructure Dbreunig.com
The era of 'free lunch' performance gains from model updates is ending, forcing developers to optimize coding harnesses and context strategies for cost-efficiency.
What: High inference costs of models like Fable are driving developers toward smaller, efficient models like GLM 5.2, which offers roughly 1/9th the cost of Fable while remaining sufficient for rote coding tasks.
Why it matters: As model performance plateaus in cost-effectiveness, the competitive advantage for developers is shifting from simply using the latest, most expensive model to building better engineering harnesses that manage context for cheaper, specialized models.
Takeaway: Invest time in refining your LLM prompt routing and context-window management instead of relying on the newest high-end model to solve all problems.
Deep dive
  • Moore's Law previously allowed developers to ignore code optimization because hardware performance doubled every 18 months.
  • Current agentic coding workflows are shifting away from 'throw it at the biggest model' due to high costs.
  • GLM 5.2 and similar models are becoming viable alternatives to premium models like Opus for routine tasks.
  • Fable's data retention and access control issues are pushing companies to rethink where they send data.
  • Better coding harnesses can provide weaker models with sufficient context to achieve parity with larger, more expensive ones.
Decoder
  • Coding Harness: A framework or software wrapper used to automate, test, and manage the interactions between an LLM and a codebase.
  • Moore's Law: The observation that the number of transistors on a microchip doubles about every two years, leading to exponential increases in computing power.
  • Inference: The process of running data through a trained machine learning model to generate a prediction or response.
Original article

Fable & The End of the Free Lunch

There’s some talk today about how agentic coders are balking at Anthropic’s pricing and adopting alternatives. I was reminded of a thought I had in the weeks following Fable’s release: the free lunch was over.

When Moore’s Law was in effect, it didn’t make sense to ruthlessly optimize your code. In 18 months, a CPU would arrive that would double your performance. Herb Sutter famously referred to this as, “the free lunch,” in a seminal essay.

When Moore’s Law slowed in the mid-2000s (specifically, single-threaded performance stagnated), we suddenly had to think about parallelization, architecture, memory locality, etc.

We had to think about what work went where.

Prior to Fable, it felt silly to waste too much time improving your coding harness or context strategies. A new model would arrive at the same price (or cheaper!) and paper over most of your problems.

But then Fable landed. It was (and still is!) incredible. But the cost was so high and Opus was good enough (as was 5.6, K3, and even GLM) for most of the code we needed.

So we started to think about what work went where.

GLM 5.2 is worth focusing on. It came out the same week as Fable and is roughly 1/9th the cost (and ~1/5th the cost of Opus 5). Is GLM 1/9th the quality of Fable? Perhaps, for certain classes of tasks. But for most rote coding it’s more than sufficient. Especially when provided with great context. I frequently chat with Fable to interrogate and shape a design, before handing off a brief to GLM.

I get pushback that falling inference prices will eventually bring us back to sending everything through the largest models. But I’m not so sure: those same gains will benefit the K3s and Qwens, and as we continue to develop better harnesses it will be easier to provide weaker (but still great) models with sufficient context to perform well.

Plus, Fable’s other shock likely locks in this change. Fable’s access controls, dynamic degradation, and required data retention spooked enough companies (and countries!) into thinking about where they send their traces and where they get their tokens.

Explore all 66 articles in today's digest

Fresh dev news, AI-digested daily. Never miss a trend at devoured.fyi