Skip to content

Budget gate for LLM spend (--max-spend) #206

Description

@fahedouch

Vens makes LLM calls and has no idea what they cost. No ceiling, no warning, nothing stops a big scan from running up the bill. On a metered API key that's a real worry, especially the first runs where you don't know the batch count yet.

Would like a budget gate: cap a run at a rough token or dollar estimate and stop (or prompt) before it blows past. Something like --max-spend or --max-tokens, checked before each batch.

A rough estimate is fine, it doesn't need to be exact. The point is a guardrail so people can try vens without fear of draining their quota.

Ref: limitations #8 (no built-in cost control).

Metadata

Metadata

Assignees

No one assigned

    Labels

    enhancementNew feature or request

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions