* High level reasons:

It increases the gap between the rich and the poor. You think indie devs and individuals have the resources to pay per token? The direction Anthropic is taking is one that leads to more inequality, only this time it's intelligence inequality between small and large corps.

Anthropic also intentionally crippled the model when it comes to doing AI research. And then Karpathy (an AI researcher who defected to Anthropic) has the audacity to praise the model?

These decisions were already made by Anthropic (a company that actively wants AI regulation and banning of open source models) BEFORE the USG request to take down the model, so they can't blame the politics for this.

* Earthly reasons:

We were originally going to get 100% usage from June 9th to June 23rd (15 days)

Instead we got 100% usage for 3 days and 50% usage for 7.

And it's nerfed on day 1:

> ... we're redeploying the model with a new set of classifiers to target and block more cybersecurity tasks. In the near term, some routine tasks like coding and debugging will fall back to Opus 4.8

If I want Fable-class of models, there's already GPT-5.5-pro (not accessible in Codex, for some reason) which I can use with my subscription. Nothing about Fable

I have yet to be able to get a single request that isn't downgraded to Opus-4.8 fairly quickly -- my most recent request (in Claude Desktop instead of Claude Code, hoping for better results) was "Please read and analyze my recent sessions, and let's discuss any themes and patterns you notice in my usage" -- somehow this was a threat.

I've found Claude and GPT have converged again. The LLMs are bound to be commoditized, in 6 months, some open-source LLM will step in.

The real problem isn't that Fable has flaws, but that a maxed-out Fable might not be something everyone gets to have equally.

It's not nerfed, it just won't respond to "fix this code" when the code is insecure and will silently fall back to a weaker model.

GLM 5.2 is already close to Fable. But none of this matters if people can't run the models on their own hardware.

Give it time and open models will get close enough.