
A mounting group of people is growing frustrated over unexpected costs connected to extended AI conversations, particularly when idle for more than 30 minutes. Recent chatter on forums show some losing up to 20% of their speaking credits due to processing issues.
This tax has users reporting significant lossesโ15-20%โwhen a conversation halts. One contributor noted, "I stopped keeping conversations alive. A fresh session is cheaper than one that starts from an idle state."
Many suspect glitches in how conversation data is managed. As one participant highlighted, "The conversation sometimes shifts to a different backend server, and we have to pay for it." When a chat times out, restarting can lead to higher processing fees.
"The whole prefix gets recomputed, which is unavoidable unless you minimize the size," a participant pointed out.
Users are increasingly vocal about the need for clearer alerts regarding potential costs. One user insisted, "There should be a warning that states if you initiate another prompt, it might lead to huge costs." This notion gained traction as many shared similar sentiments that clarity could alleviate some stress.
Another user mentioned, "I've created custom skills to save necessary context to avoid starting anew."
"For anything scheduled, I simply avoid keeping conversations alive," stressed another innovative contributor.
๐ Users report 15-20% losses tied to inactivity.
๐ Tools like Mental CLI help some manage conversation context better.
โ ๏ธ Alerts for potential costs might ease user tensions.
Ongoing dissatisfaction indicates an urgent need for transparent AI service billing. As people continue sharing experiences on forums, could this drive companies to reconsider their pricing? The community's frustrations may just be the spark for change.