Bans on AI models are worrying for everyone
Anthropic's next frontier model just got banned, this sets a terrible precedent for future models.
Comparing Fable 5 to the downgraded Opus is probably like comparing crack and sugar; it was supposedly that good. With Anthropic’s Fable already banned and now OpenAI’s GPT-5.6 Sol joining the honorary list, it’s worrying that we can get rugpulled at any time.
These bans have also set a precedent so the currently accessible models aren’t safe either. Your workflow working today may not work tomorrow if the access is taken away for any reason.
It’s also not a crazy thought that the US could restrict these companies’ offerings to certain countries due to geopolitical circumstances. For reference, it’s not possible to natively access ChatGPT, Claude, or Gemini in China.
AI is fast becoming a part of everyday life, but the more we rely on it, the more we’re at the mercy of the companies running it, who can pull the models anytime.
That said, we can’t just go back to “caveman” mode, so this brings me to some good news that we now have many capable open-weight models from Chinese and European companies that anyone can download and host.
If anything, this has created several new companies that are either renting their GPUs or hosting these models themselves with API pricing.
Open models have historically been 6+ months behind the frontier ones from Anthropic/OpenAI, but as of today’s publication date, Z.AI’s GLM 5.2 now closely rivals frontier models and, in some cases, sits just behind Claude’s banned model, Fable 5.
Benchmarks don’t mean as much anymore, but anecdotal experiences on Reddit and X seem to agree that GLM is an impressive model, with the caveat that it takes longer to respond than its competitors.
Either way, we’re starting to get powerful open-weight models that are just as strong (or even stronger) than the frontier models from AI companies, and the progress of these open-weight models not only reduces our reliance on AI companies but also genuinely contributes to humankind.
This post was made with some salt because I was just about to purchase Anthropic’s Max x20 plan, only to discover that the model is now banned…
‘Update:’ Anthropic announced good news and bad. Model will be live from July 1st but will be taken off from subscription plans after July 7th and become API-exclusive. Seeing its API pricing, you do NOT want to try that route.