" Choosing Between Codex and GLM 5.2 Based on the AI subscription services I currently have and my usage experience over the past few days, here’s what I’ve found: Codex still offers a better overall experience compared to OpenCode GO’s GLM 5.2. Two main reasons: 1. Quotas 2. Speed The GPT 5.5 on Codex has been super fast recently, and as for quotas, even a Plus account is barely sufficient but manageable. On the other hand, OpenCode GO’s GLM 5.2 is painfully slow, and the quota gets used up at lightning speed. Logically, an open-source model shouldn’t be both slow and have such low quotas. Could it be that the curse of insufficient computing power has now hit open-source models? Unless Zhipu’s official API can match DeepSeek in terms of speed and pricing, it’s hard to consider fully switching over. Maybe by then, GLM will dominate the LLM usage rankings. For now, I can only go with a hybrid approach: Codex + DeepSeek (V4 Flash + Pro) + some GLM 5.2 As for Gemini, it’s just disappointing. And Claude? I don’t even have the energy to bother.