A pseudonymous analyst tallies every piece from data acquisition through wages and lands on a $200M–$300M price tag for a frontier-scale model like GLM-5, yet a second commentator immediately flags that the sum likely sweeps in experiments unrelated to this one release.
@teortaxesTex @stevehou cost is around $200M-$300M end to end (data, compute, experiments, wages)
@stevehou @zephyr_z9 It's not Fable I can say it's around Opus 4.65 or GPT 5.4, with sparks of Opus 4.8 in small domains Also it was certainly trained on Nvidia and we don't know costs
Labs rarely run a single model in isolation, so any end-to-end figure mixes dedicated work with broader R&D; without line-item breakdowns, readers are left guessing how much of the quoted range actually belongs to GLM-5 itself.
Zhipu AI raised roughly $558 million in its January Hong Kong listing, money the company said went straight into accelerating GLM-5; that fresh runway makes the disputed cost range more than academic because it directly shaped how fast the model reached release.
Users in the replies disliked the $200M-$300M aggregate AI model training cost estimates because they suspect unrelated experiments are inflating the figures.
No Digg Deeper questions have been answered for this story yet.
@zephyr_z9 @stevehou I'll believe you but I dont like such aggregate values they must be doing experiments unrelated to GLM-5
@teortaxesTex @stevehou cost is around $200M-$300M end to end (data, compute, experiments, wages)