Skip to content

Prompt caching on Gemini + Anthropic #9

@shreydevkar

Description

@shreydevkar

Long chain-mode turns re-send the full sheet context every time. Enable prompt caching on providers that support it (Anthropic beta, Gemini CachedContent) to cut latency and cost on the 2nd-plus turn.

Acceptance criteria:

  • Provider abstraction gets a cache_id / cache_name field.
  • System-prompt + context turn gets cached on the first call.
  • Follow-up calls reference the cached prefix.

Metadata

Metadata

Assignees

No one assigned

    Labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions