Upgrade model precision to Q8_0 for better quality

- Falcon-H1 100M: Q4_K_M → Q8_0
- Gemma-3 270M: Q4_K_M → Q8_0
- ERNIE-4.5 0.3B: Q4_K_M → Q8_0
- Granite-4.0 350M: Q4_K_M → Q8_0
- LFM2 350M: Q4_K_M → Q8_0
- Hunyuan 0.5B: Q4_K_M → Q8_0
- BitCPM4 0.5B: kept at q4_0 (highest available)
- Qwen3 0.6B: kept at Q4_K_M (original precision)

Higher precision = better output quality at cost of slightly larger downloads and slower inference

app.py

@@ -33,31 +33,31 @@ AVAILABLE_MODELS = {
     "falcon_h1_100m": {
         "name": "Falcon-H1 100M",
         "repo_id": "mradermacher/Falcon-H1-Tiny-Multilingual-100M-Instruct-GGUF",
-        "filename": "*Q4_K_M.gguf",
+        "filename": "*Q8_0.gguf",
         "max_context": 32768,
     },
     "gemma3_270m": {
         "name": "Gemma-3 270M",
         "repo_id": "unsloth/gemma-3-270m-it-qat-GGUF",
-        "filename": "*Q4_K_M.gguf",
+        "filename": "*Q8_0.gguf",
         "max_context": 32768,
     },
     "ernie_300m": {
         "name": "ERNIE-4.5 0.3B (131K Context)",
         "repo_id": "unsloth/ERNIE-4.5-0.3B-PT-GGUF",
-        "filename": "*Q4_K_M.gguf",
+        "filename": "*Q8_0.gguf",
         "max_context": 131072,
     },
     "granite_350m": {
         "name": "Granite-4.0 350M",
         "repo_id": "unsloth/granite-4.0-h-350m-GGUF",
-        "filename": "*Q4_K_M.gguf",
+        "filename": "*Q8_0.gguf",
         "max_context": 32768,
     },
     "lfm2_350m": {
         "name": "LFM2 350M",
         "repo_id": "LiquidAI/LFM2-350M-GGUF",
-        "filename": "*Q4_K_M.gguf",
+        "filename": "*Q8_0.gguf",
         "max_context": 32768,
     },
     "bitcpm4_500m": {
@@ -69,7 +69,7 @@ AVAILABLE_MODELS = {
     "hunyuan_500m": {
         "name": "Hunyuan 0.5B (256K Context)",
         "repo_id": "mradermacher/Hunyuan-0.5B-Instruct-GGUF",
-        "filename": "*Q4_K_M.gguf",
+        "filename": "*Q8_0.gguf",
         "max_context": 262144,
     },
     "qwen3_600m_q4": {