最敏感前沿模型或将被限制公开发布

> the most sensitive frontier models may be banned from public release or restricted to use within China

Realistically: the CCP now has a generic playbook of "replicating every weapon the US has", particularly in export controls, as demonstrated by MOFCOM last October. They're so far been used only in retaliation. This doesn't mean V5 or something is restricted so long as it's below the best closed American model at the time. This probably does mean we're never seeing GLM 5.Cyber, that version that got the OPD from the dedicated cybersecurity expert. That's for MSS and qualified domestic companies. I think this is reasonable and inevitable. But… this IS the Europe-2031 scenario, and probably it'll be formalized before June 2027. You're living on borrowed time, guys. Pax Silica does not bind the US to keep providing you with cutting edge intelligence. Chinese private actors, subject to export controls you're collaborating with by the way, have given you near-frontier checkpoints, explained the research, open sourced advanced software like veRL/slurm/DeepSeek stack. That's pretty much their whole alpha in AI, up to some point in early 2026. Add your Blackwell commitments, and that's enough to get to Mythos at home… probably. But you'd best assume you're on your own, from now on. Help won't come.