the_contextian
Region: US
Sunday 14 June 2026 16:25:34 GMT
1117
285
4
2
Music
Download
Comments
o1dp1zza :
I've been using GLM for awhile now. It actually produces some really solid output. I've also been using it with Openclaw and have been very pleased.
2026-06-14 16:42:21
0
the_contextian :
Try GLM-5.2 on the GLM Coding Plan: z.ai/subscribe?ic=BMGXV9OQBQ Use this and get an extra 10% off!
2026-06-14 16:35:41
1
廖万里 :
To tell the truth, most of China's large models have a total parameter volume of about 1 trillion, and the largest one, DeepSeek V4 Pro, only reaches 1.6 trillion parameters. Models like GLM 5.1 only have more than 70 billion parameters, not even close to 1 trillion. However, closed-source American models such as Claude Opus 4.8 carry a total parameter volume of around 5 trillion. Furthermore, according to recently leaked internal documents, the unreleased Claude Fable 5 boasts a total parameter count of 10 trillion. Therefore, since China does not possess computing power as robust as that of the United States, its large models have consistently followed the route of achieving superior intelligence with small parameter sizes. The US owns massive computing resources, so it can ramp up model parameters without limit.
2026-06-15 14:13:23
1
To see more videos from user @the_contextian, please go to the Tikwm
homepage.
