Skip to content

The all-in-one local AI studio for your desktop: chat, image and video generation and a coding agent in one free, open source app. Windows and Linux. No Docker, no terminal, no cloud required.

merge fix/311-stable-prefix (c7a41433e): a long conversation on a paid provider keeps its prompt prefix

purple6e8326a0fe07
diff --cc CHANGELOG.md
index 3e4b7864,92a2895a..599c751d
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@@ -2,27 -2,26 +2,31 @@@
All notable changes to Locally Uncensored are documented here.
-## [3.1.1] - unreleased
+## [3.1.1] - 2026-11-17
-### Fixed
+A fix for the cost of long conversations on paid models, and a clear message
+for 3D with an older ComfyUI.
+### Fixed
- - Long conversations on paid models (LU Cloud and your own API keys) cost
- what they should again. Once a conversation was long enough to be
- shortened, LU cut it at a different place on every message. The provider
- could then never reuse what it had already read, and every step was charged
- in full: in one measured case about ten times what it should have been.
- LU now cuts on fixed marks counted from the start, so the part before the
- cut stays the same and is read from the provider's cache. Agent and Code
- runs keep the same cut from step to step.
- - To keep that cut steady, a conversation with a paid model now fills at
- most about three quarters of the model's context window (0.765 of it)
- instead of 0.8 before. Local models are not affected.
+ - Long chats and long Agent or Code runs on LU Cloud and other paid providers
- keep the provider's cache discount. Once a conversation was long enough to
- be trimmed (or had more than 380 messages), the app cut it at a different
- place on every send, so nothing could be served from the cache and a step
- cost about ten times what it should. The cut now stays in place for many
- steps and moves in whole blocks. Applies to Chat, Group, Compare, Agent and
- Code.
++ (your own API keys) keep the provider's cache discount. Once a conversation
++ was long enough to be trimmed (or had more than 380 messages), the app cut
++ it at a different place on every send, so nothing could be served from the
++ cache and a step cost about ten times what it should. The cut now stays in
++ place for many steps and moves in whole blocks. Applies to Chat, Group,
++ Compare, Agent and Code. Local models are not affected.
+- 3D in Create with a ComfyUI older than 0.42.4 says that ComfyUI needs an
+ update and offers it, instead of stopping with an error from ComfyUI's
+ RemeshMesh node.
+ ### Changed
-
+ - A paid model whose send window is set as large as the model's own window
- now sends at most about 76 percent of that window instead of 80, so the
- trim engages before the last size check has to shorten single messages.
++ now sends at most 0.765 of that window (about 76 percent) instead of 0.8,
++ so the trim engages before the last size check has to shorten single
++ messages.
+ - After a trim a request stays between 70 and 100 percent of the send budget
+ instead of between 70 and 115.
+
## [3.1.0] - 2026-11-16
LU learns to run itself from the chat and to do tasks on a schedule, the