fix(ai): add GPT-5.4 and GPT-5.5 long-context pricing
This commit is contained in:
@@ -10,6 +10,7 @@
|
||||
### Fixed
|
||||
|
||||
- Fixed post-compaction output-token budgeting to ignore stale assistant usage from before the compaction boundary ([#6464](https://github.com/earendil-works/pi/issues/6464)).
|
||||
- Fixed GPT-5.4 and GPT-5.5 long-context cost accounting while retaining the intentional 272K default context limit for models that require an explicit override.
|
||||
- Fixed GPT-5.6 metadata to keep direct OpenAI requests in the 272K short-context tier while exposing the Codex backend's 372K context window with long-context pricing.
|
||||
|
||||
## [0.80.5] - 2026-07-09
|
||||
|
||||
Reference in New Issue
Block a user