update pr review action to new model

Author: Panaetius

.github/workflows/pr-review.yaml

@@ -37,9 +37,9 @@ jobs:
         env:
           OPENAI_KEY: ${{ env.SDSC_VLLM_TOKEN }}
           GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-          config__model: "openai/Qwen/Qwen3-Coder-30B-A3B-Instruct-FP8"
-          config__fallback_models: '["openai/Qwen/Qwen3-Coder-30B-A3B-Instruct-FP8"]'
-          config__custom_model_max_tokens: "32000"
+          config__model: "openai/cyankiwi/Qwen3-Coder-30B-A3B-Instruct-AWQ-Int4"
+          config__fallback_models: '["openai/cyankiwi/Qwen3-Coder-30B-A3B-Instruct-AWQ-Int4"]'
+          config__custom_model_max_tokens: "192000"
           config__ai_timeout: "1800"
           openai__extra_body: '{"chat_template_kwargs":{"enable_thinking":false}}'
           OPENAI__API_BASE: "https://vllm-gateway-runai-sharedllm-ralf.inference.compute.datascience.ch/v1"
@@ -55,8 +55,8 @@ jobs:
       #   env:
       #     # --- LLM configuration ---
       #     LLM__PROVIDER: "OPENAI"
-      #     LLM__META__MODEL: "Qwen/Qwen3-Coder-30B-A3B-Instruct-FP8"
-      #     LLM__META__MAX_TOKENS: "16000"
+      #     LLM__META__MODEL: "cyankiwi/Qwen3-Coder-30B-A3B-Instruct-AWQ-Int4"
+      #     LLM__META__MAX_TOKENS: "192000"
       #     LLM__HTTP_CLIENT__API_URL: "https://vllm-gateway-runai-sharedllm-ralf.inference.compute.datascience.ch/v1"
       #     LLM__HTTP_CLIENT__API_TOKEN: ${{ env.SDSC_VLLM_TOKEN }}