branch: externals/llm
commit 1bd69ed78184596e3389cd115688b055888e71cb
Author: Andrew Hyatt <ahy...@gmail.com>
Commit: GitHub <nore...@github.com>
Fix bug in streaming with reasoning models (#195)
When only reasoning results have come in, we shouldn't call callbacks
with `nil` values when `multi-output` is off.
This will fix https://github.com/ahyatt/ekg/issues/203.
---
NEWS.org | 1 +
llm-provider-utils.el | 9 +++++----
2 files changed, 6 insertions(+), 4 deletions(-)
diff --git a/NEWS.org b/NEWS.org
index e531e0bc70..4d629d3d88 100644
--- a/NEWS.org
+++ b/NEWS.org
@@ -3,6 +3,7 @@
- Fix bug in ollama batch embedding generation.
- Add Qwen 3 and Gemma 3 to model list.
- Fix broken model error message
+- Fix reasoning model and streaming incompatibility
* Version 0.25.0
- Add =llm-ollama-authed= provider, which is like Ollama but takes a key.
- Set Gemini 2.5 Pro to be the default Gemini model
diff --git a/llm-provider-utils.el b/llm-provider-utils.el
index c2035d58cd..0a6d7af107 100644
--- a/llm-provider-utils.el
+++ b/llm-provider-utils.el
@@ -424,10 +424,11 @@ Any strings will be concatenated, integers will be added,
etc."
(setq current-result
(llm-provider-utils-streaming-accumulate
current-result s))
(when partial-callback
- (llm-provider-utils-callback-in-buffer
- buf partial-callback (if multi-output
- current-result
- (plist-get current-result
:text)))))
+ (when-let* ((callback-val (if multi-output
+ current-result
+ (plist-get current-result
:text))))
+ (llm-provider-utils-callback-in-buffer
+ buf partial-callback callback-val))))
(lambda (err)
(llm-provider-utils-callback-in-buffer
buf error-callback 'error
