feat: mark audio-capable models in the picker

4 files changed, 45 insertions, 7 deletions

diff --git a/CHANGELOG.md b/CHANGELOG.md
index 9eb1ee5..a2a3635 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -84,8 +84,10 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
   selected model reports audio input (`architecture.input_modalities` from the
   router, or the per-model "Accepts audio" setting); the clip is captured at
   16 kHz mono, sent as an `input_audio` content part, and discarded after
-  send, leaving only a `voice-note Ns` marker in history. Capture needs
-  QtMultimedia from PySide6-Addons and is absent without it.
+  send, leaving only a `voice-note Ns` marker in history. Audio-capable
+  models show a microphone (🎙) in the picker beside their name, as vision
+  models show an eye. Capture needs QtMultimedia from PySide6-Addons and is
+  absent without it.
 
 ### Fixed
 
diff --git a/README.md b/README.md
index b1ad066..dadc43d 100644
--- a/README.md
+++ b/README.md
@@ -519,8 +519,9 @@ short default instruction so the request always carries a text part.
 
 Capability comes from the router: llama-server's model router reports each
 model's accepted inputs in `/v1/models` (`architecture.input_modalities`), and
-a model listing `"audio"` there gets the control. Cloud models, whose
-endpoints do not report modalities, can be marked by hand with the
+a model listing `"audio"` there gets the control and a microphone (🎙) beside
+its name in the picker, the way a vision model gets an eye. Cloud models,
+whose endpoints do not report modalities, can be marked by hand with the
 **Accepts audio** box in the model settings dialog.
 
 Recording is 16 kHz mono WAV, the format speech encoders expect. The clip is
diff --git a/llamachat/ui.py b/llamachat/ui.py
index d5c6d34..91321e4 100644
--- a/llamachat/ui.py
+++ b/llamachat/ui.py
@@ -929,8 +929,7 @@ class ChatWindow(QMainWindow):
         self.model_box.clear()
         for name in available:
             info = self.model_info(name)
-            label = f"{name} 👁" if info.vision else name
-            self.model_box.addItem(label, name)
+            self.model_box.addItem(_model_label(name, info), name)
         self.model_box.blockSignals(False)
 
         target = previous or self.cfg.default_model
@@ -2305,7 +2304,17 @@ def _column(row, name: str, default: str = "") -> str:
 
 
 def _strip_marker(name: str) -> str:
-    return name.replace(" 👁", "").strip()
+    return name.replace(" 👁", "").replace(" 🎙", "").strip()
+
+
+def _model_label(name: str, info) -> str:
+    """The picker label: the model id plus a mark for each input modality.
+
+    The marks are display only; the item's data is always the bare id, and
+    `_strip_marker` is their inverse.
+    """
+    marks = (" 👁" if info.vision else "") + (" 🎙" if info.audio else "")
+    return name + marks
 
 
 def _plain(text: str) -> str:
diff --git a/test_llamachat.py b/test_llamachat.py
index 20aea57..bf43c77 100755
--- a/test_llamachat.py
+++ b/test_llamachat.py
@@ -520,6 +520,31 @@ def test_audio_bubble_marker():
     print("ok  audio bubble marker")
 
 
+def test_model_label_markers():
+    """The picker marks vision and audio, and strips both back off."""
+    import os
+
+    os.environ.setdefault("QT_QPA_PLATFORM", "offscreen")
+
+    from llamachat import models
+    from llamachat.ui import _model_label, _strip_marker
+
+    assert _model_label("m", models.ModelInfo()) == "m"
+    assert _model_label("m", models.ModelInfo(vision=True)) == "m 👁"
+    assert _model_label("m", models.ModelInfo(audio=True)) == "m 🎙"
+    assert (
+        _model_label("m", models.ModelInfo(vision=True, audio=True))
+        == "m 👁 🎙"
+    )
+
+    # The strip is the inverse, and leaves an unmarked name alone.
+    assert _strip_marker("m 👁 🎙") == "m"
+    assert _strip_marker("m 🎙") == "m"
+    assert _strip_marker("m 👁") == "m"
+    assert _strip_marker("m") == "m"
+    print("ok  model label markers")
+
+
 def test_unbalanced_backticks():
     """An odd backtick cannot reflow the rest of a reply as code."""
     import os
@@ -4065,6 +4090,7 @@ if __name__ == "__main__":
     test_prompt_column_migration()
     test_markdown_rendering()
     test_audio_bubble_marker()
+    test_model_label_markers()
     test_unbalanced_backticks()
     test_sidebar_toggle()
     test_shortcuts()