Add GLM 4.5,4.5air,4.5flash

[Shyliuli]

https://docs.z.ai/guides/llm/glm-4.5

[BradKML]

Might as well get the older versions as well (from 3 to 4) cus they are good to keep as reference

[zixuanl4]

@gnguralnick Hi Gabe, this is Z.ai Official. Just want to follow up on my email from yesterday regarding the Native Platform configuration. I know you have a very full plate, but I wanted to ensure this didn't get lost.

It appears the code is currently configured for DeepInfra inference. We are, however, still collaboratively evaluating the deployment effectiveness of the API with an external provider.

When you have a chance, would you mind reviewing the information from yesterday's email and updating the endpoint and settings? I would greatly appreciate your help with this.

[gnguralnick]

Hi @zixuanl4 , responding here too just for clarity - I've just seen your email with API information for your Native Platform and have started up our evaluation using that. I will publish the results when they are ready.

[BradKML]

@gnguralnick mind if I ask, what is the status of LiveBench in general? It seemed that the tests stopped at May? Also, how can people donate their compute so that models can be tested (e.g. if people have a lot of residual OpenRouter credits or GPU through Ollama)

• GPT's OSS line Please test GPT OSS 20b! #267
• Qwen3 Add recent Qwen3-30B-A3B models #259
• Mistral Add magistral models #235 and Add devstral #221
• others General Issue for Non-Attention models #250

[BradKML]

GLM 4.5 is alr here