Skip to content

why tokens run so quickly? #37748

Description

@1273693845

Description

The label “Kimi K3 (2x usage)” is naturally understood as offering twice the usage allowance, especially because the Go promotional page states that Kimi K3 gets “2x usage limits.” However, my dashboard suggests the opposite: $5.85 of displayed K3 cost appears to count as approximately $11.70 against the five-hour quota. Together with $0.38 from other models, this exhausted the $12 limit.

The cost chart shows only approximately $6.23, while the quota meter shows 100% of a $12 limit, without displaying any weighted usage calculation. This is materially misleading. In addition, the K3 agent stopped during a file-write step and repeatedly returned “Upstream request failed,” without completing the task. Please clarify the multiplier, provide request-level records, and restore the quota consumed by abnormal retries.

Image Image

Plugins

none

OpenCode version

the latest

Steps to reproduce

recover my usage--- i dont want totally clear to 0, but please let it goes to normal usage just 6$
what i know is that i will more carefully use k3 or other expensive LLM from now on
stop using "2x usage" just say "spend double tokens" please

Screenshot and/or share link

No response

Operating System

No response

Terminal

No response

Metadata

Metadata

Assignees

Labels

No labels
No labels

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions