OpenCode Go quota was rapidly exhausted by abnormal repeated GLM-5.2 requests with identical 38,914 input tokens #42871

Description

What happened?

my accout is: gongxiaofeng422@gmail.com

I am a newly renewed OpenCode Go user ($10/month), and I selected GLM-5.2 as my model.

Immediately after purchasing/renewing my Go subscription, I encountered abnormal usage behavior. I only performed a small number of interactions with OpenCode, but my Go usage was consumed extremely quickly.

When I checked the Usage History, I found that a single session generated a very large number of requests within approximately one minute.

The abnormal requests have the following characteristics:

Plan: OpenCode Go ($10/month)
Model: glm-5.2
Session: kcpPxZ59
Time: August 16, 2026, approximately 14:18–14:19
Input tokens: exactly 38,914 for each request
Output tokens: exactly 1 for each request
Cost: approximately $0.0101 for each request
Number of requests: a large number of repeated requests within approximately one minute

What is particularly concerning is that many consecutive requests have exactly the same input token count (38,914) and exactly the same output token count (1).

I did NOT intentionally send this number of requests. From my perspective, I only performed a few interactions with OpenCode.

This makes me suspect that OpenCode may have entered an abnormal request/retry/agent loop, or that there may be an issue between OpenCode Go and the GLM-5.2 provider.

Impact

This issue has caused my Go usage quota to be consumed extremely quickly.

I had just purchased / renewed the OpenCode Go subscription specifically to use OpenCode with GLM-5.2. However, almost immediately after purchasing/renewing the subscription, I was unable to continue using the service because my usage quota had been consumed.

This is particularly concerning because the usage appears to have been caused by repeated abnormal requests rather than by my intentional usage.

I have attached a screenshot of the Usage History showing the repeated requests.

Request for investigation

Could you please investigate:

Why did OpenCode generate such a large number of requests within approximately one minute?
Why do these requests contain exactly the same 38,914 input tokens and only 1 output token?
Was this caused by an Agent loop, retry loop, provider/API issue, or another OpenCode bug?
Why were these requests counted against my Go usage quota?
Could you please review the usage records for session kcpPxZ59 and determine which requests were abnormal?
Request for compensation

Since I had just purchased / renewed the OpenCode Go subscription and was almost immediately unable to use the service due to this abnormal behavior, I would also like to request compensation for the loss of my Go usage quota.

Please consider one of the following resolutions:

Restore the Go usage quota that was consumed by the abnormal requests; or
Provide equivalent usage credits / quota compensation; or
If the issue cannot be resolved in this way, provide an appropriate refund for the affected subscription period.

I paid for the Go subscription with the expectation that the included usage quota would be available for my normal use. I believe it would be reasonable to restore or compensate the quota that was consumed by what appears to be an abnormal system behavior.

I would appreciate it if the team could review my account and usage records and help resolve this issue as soon as possible.

Thank you.

Plugins

n/a

OpenCode version

OpenCode Desktop v1.17.13

Steps to reproduce

I have no quota now. I don't know how I can reproduce it, but it looks very abnormal.

Screenshot and/or share link

No response

Operating System

Window7

Terminal

No response