diff --git a/gallery/index.yaml b/gallery/index.yaml
index 09b686be0..f423b6596 100644
--- a/gallery/index.yaml
+++ b/gallery/index.yaml
@@ -39854,6 +39854,142 @@
     - filename: vibevoice-asr-q4_k.gguf
       uri: huggingface://cstr/vibevoice-asr-GGUF/vibevoice-asr-q4_k.gguf
       sha256: f1e87bb5c25dd469b495759e59c4554c4e8ec254f36c5c659737ff3e61ace982
+- name: vibevoice-asr-bitnet-crispasr
+  url: github:mudler/LocalAI/gallery/virtual.yaml@master
+  variants:
+    - model: vibevoice-asr-bitnet-embed-q8-crispasr
+    - model: vibevoice-asr-bitnet-vae-q5-crispasr
+    - model: vibevoice-asr-bitnet-both-q5-crispasr
+    - model: vibevoice-asr-bitnet-vae-q4-crispasr
+    - model: vibevoice-asr-bitnet-aggro-crispasr
+  urls:
+    - https://huggingface.co/microsoft/VibeVoice-ASR-BitNet
+    - https://huggingface.co/cstr/vibevoice-asr-bitnet-GGUF
+  description: |
+    Microsoft VibeVoice-ASR-BitNet is a 1.5B-parameter multilingual speech recognition model for English, Chinese, French, Italian, Korean, Portuguese, and Vietnamese. Its BitNet-trained language-model projections use ternary TQ2_0 weights. This recommended CrispASR build keeps the VAE encoder in Q8_0 and the embedding in F16, and is approximately 1.55 GB.
+  license: mit
+  tags:
+    - crispasr
+    - asr
+    - speech-recognition
+    - stt
+    - gguf
+    - bitnet
+    - multilingual
+  overrides:
+    backend: crispasr
+    known_usecases:
+      - transcript
+    name: vibevoice-asr-bitnet-crispasr
+    parameters:
+      model: vibevoice-asr-bitnet-tq2.gguf
+  files:
+    - filename: vibevoice-asr-bitnet-tq2.gguf
+      uri: huggingface://cstr/vibevoice-asr-bitnet-GGUF/vibevoice-asr-bitnet-tq2.gguf
+      sha256: 0a3053dfeffc8675673de892caa52b39ea7558091b9b422ebaeba94862a4d4a2
+- name: vibevoice-asr-bitnet-embed-q8-crispasr
+  url: github:mudler/LocalAI/gallery/virtual.yaml@master
+  urls:
+    - https://huggingface.co/microsoft/VibeVoice-ASR-BitNet
+    - https://huggingface.co/cstr/vibevoice-asr-bitnet-GGUF
+  description: |
+    Microsoft VibeVoice-ASR-BitNet is a 1.5B-parameter multilingual speech recognition model. This approximately 1.34 GB CrispASR build uses ternary TQ2_0 language-model projections, a Q8_0 VAE encoder, and a Q8_0 embedding.
+  license: mit
+  tags: [crispasr, asr, speech-recognition, stt, gguf, bitnet, multilingual]
+  overrides:
+    backend: crispasr
+    known_usecases: [transcript]
+    name: vibevoice-asr-bitnet-embed-q8-crispasr
+    parameters:
+      model: vibevoice-asr-bitnet-embed-q8.gguf
+  files:
+    - filename: vibevoice-asr-bitnet-embed-q8.gguf
+      uri: huggingface://cstr/vibevoice-asr-bitnet-GGUF/vibevoice-asr-bitnet-embed-q8.gguf
+      sha256: 450e5ae0a66a7bbe7318487a751524e4d73da799316ccffbbef0d03281bec849
+- name: vibevoice-asr-bitnet-vae-q5-crispasr
+  url: github:mudler/LocalAI/gallery/virtual.yaml@master
+  urls:
+    - https://huggingface.co/microsoft/VibeVoice-ASR-BitNet
+    - https://huggingface.co/cstr/vibevoice-asr-bitnet-GGUF
+  description: |
+    Microsoft VibeVoice-ASR-BitNet is a 1.5B-parameter multilingual speech recognition model. This approximately 1.33 GB CrispASR build uses ternary TQ2_0 language-model projections, a Q5_0 VAE encoder, and an F16 embedding.
+  license: mit
+  tags: [crispasr, asr, speech-recognition, stt, gguf, bitnet, multilingual]
+  overrides:
+    backend: crispasr
+    known_usecases: [transcript]
+    name: vibevoice-asr-bitnet-vae-q5-crispasr
+    parameters:
+      model: vibevoice-asr-bitnet-vae-q5.gguf
+  files:
+    - filename: vibevoice-asr-bitnet-vae-q5.gguf
+      uri: huggingface://cstr/vibevoice-asr-bitnet-GGUF/vibevoice-asr-bitnet-vae-q5.gguf
+      sha256: f31ec8b858ba69a425c626373e65f17bf73e383efe9a1e125f18196c6495c57f
+- name: vibevoice-asr-bitnet-both-q5-crispasr
+  url: github:mudler/LocalAI/gallery/virtual.yaml@master
+  urls:
+    - https://huggingface.co/microsoft/VibeVoice-ASR-BitNet
+    - https://huggingface.co/cstr/vibevoice-asr-bitnet-GGUF
+  description: |
+    Microsoft VibeVoice-ASR-BitNet is a 1.5B-parameter multilingual speech recognition model. This approximately 1.12 GB CrispASR build uses ternary TQ2_0 language-model projections, a Q5_0 VAE encoder, and a Q8_0 embedding.
+  license: mit
+  tags: [crispasr, asr, speech-recognition, stt, gguf, bitnet, multilingual]
+  overrides:
+    backend: crispasr
+    known_usecases: [transcript]
+    name: vibevoice-asr-bitnet-both-q5-crispasr
+    parameters:
+      model: vibevoice-asr-bitnet-both-q5.gguf
+  files:
+    - filename: vibevoice-asr-bitnet-both-q5.gguf
+      uri: huggingface://cstr/vibevoice-asr-bitnet-GGUF/vibevoice-asr-bitnet-both-q5.gguf
+      sha256: 3ad6552a214dab12786c3ae34afbbabef5701059a56a6e77cfeaf00c7cf02b80
+- name: vibevoice-asr-bitnet-vae-q4-crispasr
+  url: github:mudler/LocalAI/gallery/virtual.yaml@master
+  urls:
+    - https://huggingface.co/microsoft/VibeVoice-ASR-BitNet
+    - https://huggingface.co/cstr/vibevoice-asr-bitnet-GGUF
+  description: |
+    Microsoft VibeVoice-ASR-BitNet is a 1.5B-parameter multilingual speech recognition model. This approximately 1.26 GB CrispASR build uses ternary TQ2_0 language-model projections, a Q4_0 VAE encoder, and an F16 embedding.
+  license: mit
+  tags: [crispasr, asr, speech-recognition, stt, gguf, bitnet, multilingual]
+  overrides:
+    backend: crispasr
+    known_usecases: [transcript]
+    name: vibevoice-asr-bitnet-vae-q4-crispasr
+    parameters:
+      model: vibevoice-asr-bitnet-vae-q4.gguf
+  files:
+    - filename: vibevoice-asr-bitnet-vae-q4.gguf
+      uri: huggingface://cstr/vibevoice-asr-bitnet-GGUF/vibevoice-asr-bitnet-vae-q4.gguf
+      sha256: 72be9d760b2edb25faa8fa8a75a29d2de76837803599f3373496542ce955367b
+- name: vibevoice-asr-bitnet-aggro-crispasr
+  url: github:mudler/LocalAI/gallery/virtual.yaml@master
+  urls:
+    - https://huggingface.co/microsoft/VibeVoice-ASR-BitNet
+    - https://huggingface.co/cstr/vibevoice-asr-bitnet-GGUF
+  description: |
+    Microsoft VibeVoice-ASR-BitNet is a 1.5B-parameter multilingual speech recognition model for English, Chinese, French, Italian, Korean, Portuguese, and Vietnamese. This compact CrispASR build combines ternary TQ2_0 language-model projections with a Q4_0 VAE encoder and Q8_0 embedding, reducing the model to approximately 1.05 GB while producing the same JFK benchmark transcription as the larger builds published alongside it.
+  license: mit
+  tags:
+    - crispasr
+    - asr
+    - speech-recognition
+    - stt
+    - gguf
+    - bitnet
+    - multilingual
+  overrides:
+    backend: crispasr
+    known_usecases:
+      - transcript
+    name: vibevoice-asr-bitnet-aggro-crispasr
+    parameters:
+      model: vibevoice-asr-bitnet-aggro.gguf
+  files:
+    - filename: vibevoice-asr-bitnet-aggro.gguf
+      uri: huggingface://cstr/vibevoice-asr-bitnet-GGUF/vibevoice-asr-bitnet-aggro.gguf
+      sha256: 5f118693e984e70e33e8818921ea41b2997eec80866337610e3d14bdd5bce6d1
 - name: vibevoice-tts-crispasr
   url: github:mudler/LocalAI/gallery/virtual.yaml@master
   urls: