Add Qwen GSQ-RCO Beta 1 profile

config/install.env.example

@@ -56,6 +56,9 @@ FAST_MODEL_SHA256=54879ae8738d5938f46cb3b8cbf16bf42b8c85b7d68d7c73f062b612ec183e
 MEDIUM_MODEL_FILE=qwen-pure/qwen3.8-27b-IQ4_XS-pure.gguf
 MEDIUM_MODEL_URL=https://huggingface.co/jpetrina/Qwen3.8-27B-IQ4_XS-pure-GGUF/resolve/main/qwen3.8-27b-IQ4_XS-pure.gguf
 MEDIUM_MODEL_SHA256=ea5a3c45d407f9b9e5d2c0d647f0ea600f486f6b86b92b56d0823ba073dae675
+BETA1_MODEL_FILE=qwen3.8-27b-gsq-rco-test/Qwen3.8-27B-GSQ-RCO-IQ3_XXS-mtp.gguf
+BETA1_MODEL_URL=https://huggingface.co/ISTA-DASLab/Qwen3.8-27B-GSQ-RCO-GGUF/resolve/main/Qwen3.8-27B-GSQ-RCO-IQ3_XXS-mtp.gguf
+BETA1_MODEL_SHA256=63f29a2189a6b4cc31f81e093d3856ad293a5583114439f41ae1ed7af4093262
 LARGE_MODEL_FILE=qwen-pure/qwen3.8-27b-IQ4_XS-pure.gguf
 LARGE_MODEL_URL=https://huggingface.co/jpetrina/Qwen3.8-27B-IQ4_XS-pure-GGUF/resolve/main/qwen3.8-27b-IQ4_XS-pure.gguf
 LARGE_MODEL_SHA256=ea5a3c45d407f9b9e5d2c0d647f0ea600f486f6b86b92b56d0823ba073dae675
@@ -82,6 +85,11 @@ MEDIUM_CONTEXT=160000
 MEDIUM_BATCH_SIZE=2048
 MEDIUM_UBATCH_SIZE=128
 MEDIUM_TENSOR_SPLIT=85,15
+BETA1_CONTEXT=192000
+BETA1_BATCH_SIZE=2048
+BETA1_UBATCH_SIZE=128
+BETA1_GPU_DEVICES=GPU-8ad38c6c-5a01-9d8e-1dfa-ed662ad78fbe,GPU-4834d9d7-5b61-3004-1fb3-4ae49d482d4b
+BETA1_PARALLEL_SLOTS=1
 LARGE_CONTEXT=192000
 LARGE_BATCH_SIZE=2048
 LARGE_UBATCH_SIZE=128

config/profile-matrix.json

@@ -27,6 +27,18 @@
       "mtp": 3,
       "description": "Ausgewogenes Standardprofil für Alltag und lange agentische Aufgaben."
     },
+    {
+      "id": "beta1",
+      "alias": "qwen-beta-1",
+      "context": 192000,
+      "parallel_slots": 1,
+      "model_env": "BETA1_MODEL_FILE",
+      "model_family": "Qwen3.8-27B GSQ-RCO IQ3_XXS MTP",
+      "gpu_split": "5080 model / 3060 vision",
+      "vision": true,
+      "mtp": 3,
+      "description": "Beta 1: schnelles GSQ-RCO-Testprofil mit 192K Kontext und Vision-Projektor auf der RTX 3060."
+    },
     {
       "id": "large",
       "alias": "qwen-large",