True, but the capabilities and knowledge of that model are also frozen in time, so the value of that model declines over time.
A model that writes code without knowledge of any language or library changes for half a decade is less useful. A 2021 era chatgpt would be quite quaint in 2026.
Right now the Chinese labs might have incentives to release their models for free, and maybe Google is happy to release open weights today, but I'm sure there are already bean counters at Google salivating at the idea of having Gemini in Chrome as part of a Google AI monthly subscription just like YouTube premium and other Google subscriptions.
>True, but the capabilities and knowledge of that model are also frozen in time, so the value of that model declines over time.
Correction: The capabilities and knowledge of that model can be improved via self-distillation, so the value of that model increases over time.
This is where I think self-distillation is the main way forward, and probably the second best thing ever happened to AI/LLM after the transformer.
Based on self-distillation, the value of the open weights models will incease over time for sub-specialization through post-training and fine-tuning.
Please check these very promising recent works and results from MIT/ETH, UCLA and Apple [1],[2,[3]. For example the MIT/ETH self-distillation approach was demonstrated by a single H200 GPU. Apple approach is even simpler that it's simply called Simple Self-Distillation (SSD), pun intended.
[1] Self-Distillation Enables Continual Learning:
https://arxiv.org/abs/2601.19897
[2] Self-Distilled Reasoner: On-Policy Self-Distillation for Large Language Models:
https://arxiv.org/abs/2601.18734
[3] Embarrassingly Simple Self-Distillation Improves Code Generation:
https://arxiv.org/abs/2604.01193
> capabilities and knowledge of that model are also frozen in time
I think this matters less than you think. If the spigot turns off, open LLM research is going to have a powerful incentive to focus on post-training to refresh stale base models. And post-training, in general, is so much cheaper and faster than pre-training anyway. I was pretty surprised to learn that GLM-5.2's entire RL training (the part that makes it reliable at agentic tasks) was completed in just TWO DAYS.
If the world ends all I have to do is power my desktop and I'll have my locals - a decent iteration of Deepseek and a few smaller models, some focused, some just older versions - having several is key tho. They can be cross referenced to limit hallucinations and inaccurate information - this means I can confidently say that I have on my desktop - all of human history, knowledge, discoveries, maths, languages - at least in summary or truncated form (also another bonus of multiple models - will often have more comprehensive total output than one model provides) and all of those models have absolutely no restrictions other than the broadest limits allowed by current laws - so, practically no limits (I bet I could get them all it to explain splitting the atom with minor effort).
I realize that my amazing tool/system of local AI is out of date - I still very much like having it and it is not at all a bad thing to hav. Everyone in theory ought to have a local backup - for just in case.
The fact that people will have this in this one, albeit extreme, example - it would most definitely matter in the event of a societal collapse. Not everyone will have it - can they run those giant data centers off a few solar panels like a desktop PC?
For this one existential reason alone, I recommend everyone at least play around local enough to have a few models functional.
Fine tuning and updating is far cheaper than training from scratch.
The weights are not frozen in time. You can train the model on new data. It's just a matter of economics of whether you have a leading lab pay for the training or you pay for it. For the past few years having the labs do it has been the economical choice but if they stop doing so the choice will shift back to the users.