feat: 更新主流大模型供应商目录

- 按云服务、Coding Plan、API 网关和本地模型分组

- 新增兼容模型发现与供应商协议校验

frontend/src/locales/en-US.json

@@ -8380,12 +8380,14 @@
       "eventAborted": "Task aborted",
       "eventFailed": "Task execution failed"
     },
-    "cloudService": "Cloud Services",
+    "agentCliSharedConfig": "This connection is shared by AI Assistant and Agent CLI",
+    "apiGateway": "API Gateways",
+    "cloudService": "Official Cloud Services",
     "codingPlan": "Coding Plan",
     "collapseHistory": "Collapse history",
     "confidenceLabel": "Confidence",
     "confidencePercent": "Confidence {{n}}%",
-    "configHelpText": "Cloud Services: Select a provider → Activate and enter API Key → Test Connection → Save.\n          Coding Plans: GLM / Kimi / MiniMax fixed monthly plans, ideal for frequent Agent CLI usage (note: keys are not shared with pay-as-you-go versions).\n          Custom LLM: Enter endpoint + API Key + model name, compatible with OpenRouter and other OpenAI-format services.\n          Local/Self-hosted (Ollama, vLLM): Enter endpoint and model name; API Key can be left blank.\n          Testing sends \"What model are you? Please answer briefly.\" to confirm the target model.",
+    "configHelpText": "Official cloud service: Select a provider and enter its API Key; you can load the models available to that key.\n          API gateway: For New API / One API, enter the site's /v1 URL, token, and model name.\n          Local/self-hosted (Ollama, vLLM, LM Studio): Enter the Endpoint and model name; API Key is optional.\n          Test the connection before saving; the test sends one short message to confirm the actual model.",
     "configLlmFirst": "Please configure an LLM first...",
     "confirmDeleteSession": "Delete \"{{title}}\"? This action cannot be undone.",
     "confirmExecute": "Confirm Execute",
@@ -8406,7 +8408,7 @@
     "currentValue": "Current",
     "customAccess": "Custom Access",
     "customProvider": "Custom LLM",
-    "customProviderDesc": "Fill in an OpenAI-compatible endpoint, API Key (blank if no auth) and model name; works with OpenRouter, intranet proxies and other unlisted services",
+    "customProviderDesc": "Enter an OpenAI-compatible endpoint, API Key (blank if no auth), and model name; suitable for internal proxies and other unlisted services",
     "customProxyHint": "Direct access from China may be restricted; use your own network or a compatible proxy via \"Custom LLM\"",
     "dailyOpsReport": "Generate ops report",
     "daysToThreshold": "reaches threshold in {{days}} days",
@@ -8420,6 +8422,7 @@
     "enableLlm": "Enable LLM Service",
     "enableReasoning": "Enable reasoning mode",
     "endpointLabel": "Endpoint",
+    "endpointModelHint": "Enter the endpoint ID created in the provider console",
     "errorOccurred": "An error occurred",
     "examplePrefix": "Ex: ",
     "executeAutoFix": "Execute Auto-Fix",
@@ -8479,12 +8482,14 @@
     "llmTopP": "Top P",
     "llmVisionModel": "Vision Model",
     "loadAlertsDot": "Loading alert data...",
+    "loadModels": "Load Models",
     "loadAnomalyDot": "Loading anomaly forecast...",
     "loadingCapacityData": "Loading capacity data...",
     "loadingMetricData": "Loading metric data...",
     "localNoKeyHint": "Local or intranet self-hosted services usually need no key; fill it in only if the gateway requires auth.",
     "localSelfHosted": "Local / Self-hosted",
     "manualInputModel": "Enter model name manually",
+    "manualModelHint": "Enter the model ID shown in the provider console",
     "mayRecur": "{{name}} may recur",
     "mayTrigger": "{{name}} may trigger",
     "memoryKeyword": "Memory",
@@ -8495,11 +8500,14 @@
     "minimaxPayg": "MiniMax (pay-as-you-go)",
     "modelColon": "Model: ",
     "modelLabel": "Model",
+    "modelListFailed": "Could not load models; check the Endpoint, API Key, and upstream service",
     "modelPlaceholder": "e.g. deepseek/deepseek-chat or gpt-4o",
     "modelUnknown": "Unknown",
+    "modelsDiscovered": "Loaded {{count}} models",
     "networkKeyword": "Network",
     "newConversation": "New Chat",
     "newConversationStarted": "New conversation started",
+    "newApiHint": "Enter the OpenAI-compatible URL of your New API or One API instance (usually ending in /v1), its token, and the model name",
     "newSessionBtn": "New",
     "newTask": "New Task",
     "next12Hours": "Next 12 hours",
@@ -8515,6 +8523,7 @@
     "noHostsHint": "No hosts yet; add servers in \"Asset Management\" first",
     "noMetricData": "No metric data",
     "noModelContent": "(The model returned no content)",
+    "noModelsDiscovered": "The upstream returned no models; enter the model name manually",
     "noneFallback": "(none)",
     "notConnectedHint": "Connection not ready; check the LLM config or refresh the page",
     "notConnectedRetry": "Connection not ready; please retry later",
@@ -8620,23 +8629,33 @@
     "providerAndCredential": "Model & Credential",
     "providerAnthropic": "Anthropic",
     "providerAzure": "Azure OpenAI",
+    "providerBaidu": "Baidu Qianfan",
     "providerBaichuan": "Baichuan",
+    "providerCerebras": "Cerebras",
     "providerCohere": "Cohere",
     "providerColon": "Provider: ",
     "providerCustom": "Custom",
     "providerDeepseek": "DeepSeek",
     "providerGemini": "Google Gemini",
+    "providerGroq": "Groq",
     "providerLabel": "Provider",
+    "providerLmStudio": "LM Studio",
     "providerMinimax": "MiniMax",
     "providerMistral": "Mistral",
     "providerMoonshot": "Moonshot",
+    "providerNewApi": "New API / One API",
     "providerNeedsKey": "To use {{name}}, you need to activate the service and obtain an API Key first",
     "providerOllama": "Ollama",
     "providerOpenai": "OpenAI",
     "providerOpenrouter": "OpenRouter",
+    "providerPerplexity": "Perplexity",
     "providerQwen": "Qwen",
     "providerSiliconflow": "SiliconFlow",
     "providerSpark": "iFlytek Spark",
+    "providerTencentTokenHub": "Tencent Cloud TokenHub",
+    "providerTogether": "Together AI",
+    "providerVolcengine": "Volcengine Ark",
+    "providerXai": "xAI Grok",
     "providerYi": "01.AI",
     "providerZhipu": "Zhipu GLM",
     "quickActions": "Quick Actions",
@@ -8646,7 +8665,7 @@
     "reasoningHigh": "High",
     "reasoningLow": "Low",
     "reasoningMedium": "Medium",
-    "reasoningModelHint": "Suitable for reasoning models like OpenAI o-series / DeepSeek R1 / Claude 3.7 Sonnet. May cause errors with standard models.",
+    "reasoningModelHint": "For reasoning models such as OpenAI GPT-5 / o-series, DeepSeek R1, and Claude Sonnet / Opus 4.",
     "reasoningOn": "Reasoning mode is on",
     "reasoningPlanMode": "Reasoning / Plan Mode",
     "recommended": "Recommended",

frontend/src/locales/zh-CN.json

@@ -8380,12 +8380,14 @@
       "eventAborted": "任务已取消",
       "eventFailed": "任务执行失败"
     },
-    "cloudService": "云服务",
+    "agentCliSharedConfig": "智能助手与 Agent CLI 共用此连接配置",
+    "apiGateway": "API 网关",
+    "cloudService": "官方云服务",
     "codingPlan": "编码套餐 Coding Plan",
     "collapseHistory": "收起历史任务",
     "confidenceLabel": "置信度",
     "confidencePercent": "置信度 {{n}}%",
-    "configHelpText": "云服务：选择提供商 → 开通并填写 API Key → 测试连接 → 保存。\n          编码套餐：GLM / Kimi / MiniMax 固定月费套餐，适合 Agent CLI 高频调用（注意 Key 与按量版不互通）。\n          大模型自定义：填请求地址 + API Key + 模型名，兼容 OpenRouter 等 OpenAI 格式服务。\n          本地/自建（Ollama、vLLM）：填写 Endpoint 与模型名即可，API Key 可留空。\n          测试时会发送「你是什么模型？请简短回答。」以确认接入目标模型。",
+    "configHelpText": "官方云服务：选择提供商并填写 API Key，可读取当前密钥可用的模型。\n          API 网关：New API / One API 需填写站点的 /v1 地址、令牌和模型名。\n          本地/自建（Ollama、vLLM、LM Studio）：填写 Endpoint 与模型名即可，API Key 可留空。\n          保存前必须先测试连接；测试会发送一条简短消息确认实际接入模型。",
     "configLlmFirst": "请先配置大模型...",
     "confirmDeleteSession": "确定删除「{{title}}」？此操作不可恢复。",
     "confirmExecute": "确认执行",
@@ -8406,7 +8408,7 @@
     "currentValue": "当前值",
     "customAccess": "自定义接入",
     "customProvider": "大模型自定义",
-    "customProviderDesc": "填写 OpenAI 兼容的请求地址、API Key（无鉴权可留空）和模型名；适用于 OpenRouter、内网代理、其他未列出的服务",
+    "customProviderDesc": "填写 OpenAI 兼容的请求地址、API Key（无鉴权可留空）和模型名；适用于内网代理或其他未列出的服务",
     "customProxyHint": "国内直连可能受限，需自备网络或使用「大模型自定义」接入兼容代理",
     "dailyOpsReport": "生成运维报告",
     "daysToThreshold": "{{days}} 天后触及阈值",
@@ -8420,6 +8422,7 @@
     "enableLlm": "启用 LLM 服务",
     "enableReasoning": "启用推理模式",
     "endpointLabel": "接口地址（Endpoint）",
+    "endpointModelHint": "请填写在提供商控制台创建的接入点 ID",
     "errorOccurred": "发生错误",
     "examplePrefix": "例：",
     "executeAutoFix": "执行自动修复",
@@ -8479,12 +8482,14 @@
     "llmTopP": "Top P",
     "llmVisionModel": "视觉模型",
     "loadAlertsDot": "加载告警数据...",
+    "loadModels": "获取模型",
     "loadAnomalyDot": "加载异常预测...",
     "loadingCapacityData": "加载容量数据...",
     "loadingMetricData": "加载指标数据...",
     "localNoKeyHint": "本地或内网自建通常无需密钥；若网关启用了鉴权再填写。",
     "localSelfHosted": "本地 / 自建",
     "manualInputModel": "手动输入模型名",
+    "manualModelHint": "请按提供商控制台显示的模型 ID 手动填写",
     "mayRecur": "{{name}}可能再次发生",
     "mayTrigger": "{{name}}可能触发",
     "memoryKeyword": "内存",
@@ -8495,11 +8500,14 @@
     "minimaxPayg": "MiniMax (按量)",
     "modelColon": "模型:",
     "modelLabel": "模型",
+    "modelListFailed": "读取模型列表失败，请检查 Endpoint、API Key 和上游服务",
     "modelPlaceholder": "如 deepseek/deepseek-chat 或 gpt-4o",
     "modelUnknown": "未知",
+    "modelsDiscovered": "已加载 {{count}} 个模型",
     "networkKeyword": "网络",
     "newConversation": "新对话",
     "newConversationStarted": "已开始新对话",
+    "newApiHint": "填写 New API 或 One API 站点的 OpenAI 兼容地址（通常以 /v1 结尾）、令牌和模型名",
     "newSessionBtn": "新建",
     "newTask": "新建任务",
     "next12Hours": "未来12小时",
@@ -8515,6 +8523,7 @@
     "noHostsHint": "暂无主机，请先前往「资产管理」添加服务器",
     "noMetricData": "暂无指标数据",
     "noModelContent": "（模型未返回内容）",
+    "noModelsDiscovered": "上游没有返回可用模型，请手动填写模型名",
     "noneFallback": "（无）",
     "notConnectedHint": "连接未就绪，请检查大模型配置或刷新页面重试",
     "notConnectedRetry": "连接未就绪，请稍后重试",
@@ -8620,23 +8629,33 @@
     "providerAndCredential": "模型与凭证",
     "providerAnthropic": "Anthropic",
     "providerAzure": "Azure OpenAI",
+    "providerBaidu": "百度千帆",
     "providerBaichuan": "百川",
+    "providerCerebras": "Cerebras",
     "providerCohere": "Cohere",
     "providerColon": "提供商:",
     "providerCustom": "自定义",
     "providerDeepseek": "DeepSeek",
     "providerGemini": "Google Gemini",
+    "providerGroq": "Groq",
     "providerLabel": "提供商",
+    "providerLmStudio": "LM Studio",
     "providerMinimax": "MiniMax",
     "providerMistral": "Mistral",
     "providerMoonshot": "月之暗面",
+    "providerNewApi": "New API / One API",
     "providerNeedsKey": "使用 {{name}} 需先开通服务并获取 API Key",
     "providerOllama": "Ollama",
     "providerOpenai": "OpenAI",
     "providerOpenrouter": "OpenRouter",
+    "providerPerplexity": "Perplexity",
     "providerQwen": "通义千问",
     "providerSiliconflow": "硅基流动",
     "providerSpark": "讯飞星火",
+    "providerTencentTokenHub": "腾讯云 TokenHub",
+    "providerTogether": "Together AI",
+    "providerVolcengine": "火山引擎方舟",
+    "providerXai": "xAI Grok",
     "providerYi": "零一万物",
     "providerZhipu": "智谱 GLM",
     "quickActions": "快速操作",
@@ -8646,7 +8665,7 @@
     "reasoningHigh": "高",
     "reasoningLow": "低",
     "reasoningMedium": "中",
-    "reasoningModelHint": "适用于 OpenAI o-series / DeepSeek R1 / Claude 3.7 Sonnet 等推理模型；普通模型开启后可能报错。",
+    "reasoningModelHint": "适用于 OpenAI GPT-5 / o-series、DeepSeek R1、Claude Sonnet / Opus 4 等推理模型。",
     "reasoningOn": "推理模式已开启",
     "reasoningPlanMode": "推理 / Plan 模式",
     "recommended": "推荐",