Initial model card — Gemma 4 family

README.md

@@ -0,0 +1,46 @@
+---
+license: apache-2.0
+tags:
+  - ava
+  - voiceos
+  - gemma
+  - on-device
+  - mixture-of-experts
+base_model:
+  - google/gemma-4-E2B-it
+  - google/gemma-4-E4B-it
+  - google/gemma-4-26B-A4B-it
+  - google/gemma-4-31B-it
+---
+
+# AVA Storm Models (Gemma 4)
+
+On-device LLM models for VoiceOS, based on Google Gemma 4.
+
+## Models
+
+| AVA ID | Source | Active Params | Format | Target |
+|--------|--------|--------------|--------|--------|
+| AVA-STORM-2B | gemma-4-E2B-it | 2B | safetensors + GGUF | Phone (low-end) |
+| AVA-STORM-4B | gemma-4-E4B-it | 4B | safetensors + GGUF | Phone (mid-range) |
+| AVA-STORM-26B-A4B | gemma-4-26B-A4B-it | 4B active (MoE) | safetensors + GGUF | Phone (flagship) / HMT-1 |
+| AVA-STORM-31B | gemma-4-31B-it | 31B | safetensors | Desktop |
+
+## Directory Structure
+
+```
+raw/                    # Base model weights (safetensors)
+  gemma-4-E2B-it/
+  gemma-4-E4B-it/
+  gemma-4-26B-A4B-it/
+  gemma-4-31B-it/
+gguf/                   # Quantized for on-device inference
+  gemma-4-E2B-it-Q4_K_M.gguf
+  gemma-4-E4B-it-Q4_K_M.gguf
+  gemma-4-26B-A4B-it-Q4_K_M.gguf
+production/             # AON-encrypted (deployed to devices)
+```
+
+## License
+
+Raw models: Apache 2.0 (Google). AON-encrypted production variants: Proprietary (Intelligent Devices LLC).