update qwen3-coder-480b-a35b-instruct-fp8 metadata

Author: nicole-lihui

models/alibaba/qwen3-coder-480b-a35b-instruct-fp8-/metadata.yaml

@@ -0,0 +1,56 @@
+apiVersion: model.hydra.io/v1alpha1
+kind: ModelSpec
+metadata:
+  name: qwen3-coder-480b-a35b-instruct-fp8
+spec:
+  config:
+    maxTokens: 262144
+  deployments:
+  - customRuntimeArgs:
+      - --max-model-len=131072
+      - --enable-auto-tool-choice
+      - --tool-call-parser
+      - qwen3_coder
+    resourceRequirements:
+      cpu: 16
+      gpuCount: 8
+      gpuType: vgpu
+      memory: 640
+      perGPUMemoryGB: 80
+    runtime: vllm
+    versionRequired: '>=0.8.5'
+  descriptor:
+    description:
+      enUS: 'Qwen3-Coder is the latest generation of agentic code models in the 
+        Qwen series, designed for advanced coding, reasoning, and agentic tasks. 
+        The flagship model, Qwen3-Coder-480B-A35B-Instruct, delivers state-of-the-art 
+        performance across Agentic Coding, Browser-Use, and fundamental programming 
+        benchmarks, matching Claude Sonnet in capability. It supports long-context 
+        understanding with native 256K tokens and extension to 1M via Yarn, 
+        optimized for large-scale repository comprehension, and features robust 
+        agentic coding support across platforms like Qwen Code and CLINE with a 
+        specialized function call design'
+      zhCN: 'Qwen3-Coder 是通义系列中最新一代具备智能体能力的代码模型，专为高级编程、推理和智能体任务设计。
+        旗舰模型 Qwen3-Coder-480B-A35B-Instruct 在智能体编程、浏览任务及基础代码基准测试中表现卓越，
+        可与 Claude Sonnet 相媲美。其原生支持 256K Token 长上下文，并可通过 Yarn 扩展至 1M，
+        优化用于仓库级代码理解；同时在 Qwen Code、CLINE 等平台上提供完善的智能体编程支持，具备专门设计的函数调用格式'
+    display: Qwen3-Coder-480B-A35B-Instruct-FP8
+    icon:
+      src: https://public-resources.d.run/models/logos/qwen-model-logo.svg
+      type: image/svg
+    links:
+    - description: About
+      url: https://github.com/QwenLM
+    provider:
+      id: alibaba
+      name:
+        enUS: Alibaba
+        zhCN: 通义千问
+    tags:
+    - TEXT_GENERATION
+    - TOOLS
+  source:
+    huggingface:
+      name: Qwen/Qwen3-Coder-480B-A35B-Instruct-FP8
+    modelscope:
+      name: Qwen/Qwen3-Coder-480B-A35B-Instruct-FP8