Skip to content

Search Older Matching Chats

Search Older Matching Chats is basically Reuse Matching Chat with a better memory.

Normally, Reuse Matching Chat only remembers the very last prompt/chat it used. With Search Older Matching Chats enabled, IntenseRP can keep up to 7 older cached chats around and reuse one of those if the current request matches.

It is off by default, and it only works when Reuse Matching Chat is also enabled for that provider.

Supported providers

Right now this feature works on:

  • DeepSeek
  • GLM Chat
  • Moonshot / Kimi
  • QwenLM
  • HuggingChat

It does not work on Perplexity or Google AI Studio right now.

How it works

  1. IntenseRP still tries the normal Reuse Matching Chat flow first.
  2. If the current chat is already the right one, it just presses Regenerate there like usual.
  3. If not, IntenseRP checks the older cached slots for the same prompt + relevant provider state.
  4. If it finds a match, it opens that cached chat URL and tries to regenerate there instead.
  5. If the regenerate button is missing, that cache entry gets dropped and IntenseRP falls back to sending the request normally.

The cache keeps the last 7 saved chats per provider/account. When a new one gets added, the oldest one is removed.

If an old cached chat gets reused successfully, it stays where it is. IntenseRP does not bump it to the top or reshuffle the cache order.

What counts as a match?

It is not just the prompt text.

IntenseRP also checks the relevant provider-side request state, such as things like:

  • the currently selected model (where that provider exposes one)
  • whether thinking is enabled
  • whether search is enabled
  • whether the prompt was sent as a text file
  • provider-specific model extras, such as HuggingChat inference provider / thinking effort

That way it does not accidentally reopen an old chat and regenerate it under the wrong setup.

Cache scope

Search Older Matching Chats is scoped per provider and per account.

That matters because provider chat IDs are usually only valid for the account that created them. A DeepSeek chat from Account A is not something Account B can normally reopen and regenerate.

Manual-login fallback

When IntenseRP knows which saved account is active, it separates the cache by account email.

If it cannot know that cleanly (for example a manual-login flow without a saved account), it falls back to the active browser profile instead. That is the safest approximation available across all providers.

DeepSeek special case

DeepSeek has one extra rule here.

If a DeepSeek chat gets content-filtered / censored, IntenseRP does not save that chat into the older-chat cache. DeepSeek disables regeneration on censored chats anyway, so keeping those around would just create dead cache entries.

Where to enable it

You enable it in the provider Behavior pages:

  • Provider Behavior -> DeepSeek -> Search Older Matching Chats
  • Provider Behavior -> GLM Chat -> Search Older Matching Chats
  • Provider Behavior -> Moonshot -> Search Older Matching Chats
  • Provider Behavior -> QwenLM -> Search Older Matching Chats
  • Provider Behavior -> HuggingChat -> Search Older Matching Chats

And again, it only does anything if Reuse Matching Chat is already enabled for that provider.

Back to Features

Features