diff --git a/mateclaw-server/src/main/java/vip/mate/agent/AgentGraphBuilder.java b/mateclaw-server/src/main/java/vip/mate/agent/AgentGraphBuilder.java
index 45d98dd4..965b3099 100644
--- a/mateclaw-server/src/main/java/vip/mate/agent/AgentGraphBuilder.java
+++ b/mateclaw-server/src/main/java/vip/mate/agent/AgentGraphBuilder.java
@@ -1408,17 +1408,28 @@ public class AgentGraphBuilder {
         return result;
     }
 
+    // Trailing "/v{digits}" segment in a base URL — the OpenAI-compatible convention
+    // (/v1 OpenAI, /v3 Volcano Ark, /v4 Zhipu). When the baseUrl already carries this
+    // segment, the default /v1 prefix on the path must be stripped to avoid building
+    // a broken URL like /api/v3/v1/chat/completions.
+    private static final java.util.regex.Pattern OPENAI_BASE_URL_VERSION_SUFFIX =
+            java.util.regex.Pattern.compile(".*/v\\d+$");
+
     private String resolveOpenAiCompletionsPath(String baseUrl, Map<String, Object> kwargs) {
         Object raw = kwargs.get("completionsPath");
-        String path = raw instanceof String value && StringUtils.hasText(value) ? value.trim() : "/v1/chat/completions";
+        boolean explicit = raw instanceof String value && StringUtils.hasText(value);
+        String path = explicit ? ((String) raw).trim() : "/v1/chat/completions";
         if (!path.startsWith("/")) {
             path = "/" + path;
         }
-        if (baseUrl.endsWith("/v1") && path.startsWith("/v1/")) {
+        // An explicit completionsPath is honored as-is. Otherwise, dedupe the /v1
+        // prefix when the baseUrl already ends with /v{N} (Volcano Engine Ark /v3,
+        // Zhipu /v4, etc.).
+        if (!explicit
+                && baseUrl != null
+                && OPENAI_BASE_URL_VERSION_SUFFIX.matcher(baseUrl).matches()
+                && path.startsWith("/v1/")) {
             path = path.substring(3);
-            if (!path.startsWith("/")) {
-                path = "/" + path;
-            }
         }
         return path;
     }
diff --git a/mateclaw-server/src/main/java/vip/mate/agent/graph/NodeStreamingChatHelper.java b/mateclaw-server/src/main/java/vip/mate/agent/graph/NodeStreamingChatHelper.java
index a852cd14..c222b8a8 100644
--- a/mateclaw-server/src/main/java/vip/mate/agent/graph/NodeStreamingChatHelper.java
+++ b/mateclaw-server/src/main/java/vip/mate/agent/graph/NodeStreamingChatHelper.java
@@ -349,7 +349,12 @@ public class NodeStreamingChatHelper {
                 || msg.contains("does not exist")
                 || msg.contains("[InvalidParameter]")
                 || msg.contains("InvalidParameter")
-                || msg.contains("url error")) {
+                || msg.contains("url error")
+                // Volcano Ark: model exists but the user's account hasn't opened it,
+                // or the id isn't valid for this region. Both are hard failures —
+                // retrying won't help, and a different provider may serve the model.
+                || msg.contains("ModelNotOpen")
+                || msg.contains("InvalidEndpointOrModel")) {
             return ErrorType.MODEL_NOT_FOUND;
         }
         // Client errors (400 Bad Request — unsupported format, invalid params, etc.) — NOT retryable
@@ -376,6 +381,20 @@ public class NodeStreamingChatHelper {
                 sb.append(cur.getMessage()).append(" | ");
             }
             sb.append(cur.getClass().getSimpleName()).append(" | ");
+            // Include the HTTP response body for WebClient errors. Many providers
+            // (Volcano Ark, Ollama, …) put the actionable error code only in the
+            // body, while the surface message is just "404 Not Found from POST X".
+            // Without this, classifyError() can never see codes like ModelNotOpen.
+            if (cur instanceof WebClientResponseException wre) {
+                try {
+                    String body = wre.getResponseBodyAsString();
+                    if (body != null && !body.isEmpty()) {
+                        sb.append(body.length() > 1024 ? body.substring(0, 1024) : body)
+                          .append(" | ");
+                    }
+                } catch (Exception ignored) {
+                }
+            }
             cur = cur.getCause();
         }
         return sb.toString();
@@ -1155,6 +1174,19 @@ public class NodeStreamingChatHelper {
         }
     }
 
+    // Pulls the offending model id out of a Volcano Ark error body. Both
+    // ModelNotOpen and InvalidEndpointOrModel.NotFound mention it after a
+    // recognizable phrase ("activated the model X" / "model or endpoint X").
+    private static final java.util.regex.Pattern ARK_MODEL_NAME_PATTERN =
+            java.util.regex.Pattern.compile(
+                    "(?:activated the model|model or endpoint)\\s+([A-Za-z0-9._-]+)");
+
+    private static String extractArkModelName(String body) {
+        if (body == null) return null;
+        java.util.regex.Matcher m = ARK_MODEL_NAME_PATTERN.matcher(body);
+        return m.find() ? m.group(1) : null;
+    }
+
     /** 从异常链提取用户友好的错误信息 */
     private static String extractUserFriendlyError(Throwable error) {
         String msg = error.getMessage();
@@ -1188,6 +1220,26 @@ public class NodeStreamingChatHelper {
                     + "例如 qwen3、qwen2.5:7b+、llama3.1:8b+、mistral-nemo、command-r 等。";
         }
 
+        // Volcano Engine Ark — model exists but the user's account hasn't activated it.
+        // Body shape: {"error":{"code":"ModelNotOpen","message":"Your account ... has not activated the model X. Please activate the model service in the Ark Console..."}}
+        if (combined.contains("ModelNotOpen")) {
+            String modelId = extractArkModelName(combined);
+            String suffix = modelId != null ? "「" + modelId + "」" : "";
+            return "火山方舟（Volcano Ark）尚未为该账号开通模型" + suffix
+                    + "。请前往 Ark 控制台 → 模型广场，对该模型点击「开通服务」后重试。"
+                    + "（控制台：https://console.volcengine.com/ark）";
+        }
+
+        // Volcano Engine Ark — model id doesn't exist for the user's region/key.
+        // Body shape: {"error":{"code":"InvalidEndpointOrModel.NotFound","message":"The model or endpoint X does not exist or you do not have access to it..."}}
+        if (combined.contains("InvalidEndpointOrModel")) {
+            String modelId = extractArkModelName(combined);
+            String suffix = modelId != null ? "「" + modelId + "」" : "";
+            return "火山方舟（Volcano Ark）找不到模型" + suffix
+                    + "。原因可能是模型 ID 不在当前区域，或你的账号没有访问权限。"
+                    + "建议在 设置 → 模型 里点「刷新模型」重新发现，或在 Ark 控制台创建「推理接入点」(ep-XXX) 后使用该 ID。";
+        }
+
         // DashScope "url error" is really "model name not mapped to any valid endpoint".
         if (msg.contains("url error") || msg.contains("[InvalidParameter]")
                 || msg.contains("Model not exist") || msg.contains("model_not_found")
diff --git a/mateclaw-server/src/main/java/vip/mate/llm/service/ModelDiscoveryService.java b/mateclaw-server/src/main/java/vip/mate/llm/service/ModelDiscoveryService.java
index e43ffe20..417fb6dc 100644
--- a/mateclaw-server/src/main/java/vip/mate/llm/service/ModelDiscoveryService.java
+++ b/mateclaw-server/src/main/java/vip/mate/llm/service/ModelDiscoveryService.java
@@ -385,11 +385,11 @@ public class ModelDiscoveryService {
                 .defaultHeader(HttpHeaders.ACCEPT, MediaType.APPLICATION_JSON_VALUE)
                 .build();
 
-        RestClient.RequestHeadersSpec<?> spec = client.get().uri("/v1/models");
+        RestClient.RequestHeadersSpec<?> spec = client.get().uri(resolveModelsPath(baseUrl));
         if (modelProviderService.hasUsableApiKey(apiKey)) {
             spec = spec.header(HttpHeaders.AUTHORIZATION, "Bearer " + apiKey.trim());
         }
-        // 添加自定义 headers（从 generateKwargs 中读取）
+        // Apply any custom headers declared in generateKwargs.
         Map<String, Object> kwargs = modelProviderService.readProviderGenerateKwargs(provider);
         applyCustomHeaders(spec, kwargs);
 
@@ -717,29 +717,51 @@ public class ModelDiscoveryService {
 
     // ==================== 工具方法 ====================
 
+    // Trailing "/v{digits}" segment in a base URL — restricted to numeric major versions,
+    // which is the OpenAI-compatible convention (/v1 OpenAI, /v3 Volcano Ark, /v4 Zhipu).
+    private static final java.util.regex.Pattern BASE_URL_VERSION_SUFFIX =
+            java.util.regex.Pattern.compile(".*/v\\d+$");
+
     /**
-     * 从 generateKwargs 中解析 completionsPath，处理 baseUrl 与路径前缀的重叠。
-     * 例如：baseUrl 以 /v4 结尾，completionsPath 为 /chat/completions → 最终 /chat/completions
-     *       baseUrl 以 /v1 结尾，completionsPath 为 /v1/chat/completions → 最终 /chat/completions
+     * Resolve the chat-completions path. An explicit {@code completionsPath} in
+     * {@code generateKwargs} is always honored as-is. Otherwise we default to
+     * {@code /v1/chat/completions} and dedupe the {@code /v1} prefix when the
+     * baseUrl already carries a {@code /v{N}} segment (e.g. Volcano Engine Ark
+     * base {@code https://ark.cn-beijing.volces.com/api/v3}).
      */
     private String resolveCompletionsPath(String baseUrl, Map<String, Object> kwargs) {
-        String path = "/v1/chat/completions";
         if (kwargs != null) {
             Object raw = kwargs.get("completionsPath");
             if (raw instanceof String value && StringUtils.hasText(value)) {
-                path = value.trim();
+                String path = value.trim();
                 if (!path.startsWith("/")) {
                     path = "/" + path;
                 }
+                return path;
             }
         }
-        // 避免路径重叠：如果 baseUrl 以 /v1 结尾且 path 以 /v1/ 开头，去掉重复
-        if (baseUrl != null && baseUrl.endsWith("/v1") && path.startsWith("/v1/")) {
+        String path = "/v1/chat/completions";
+        // If baseUrl already ends with /v{N}, strip the /v1 prefix from the default
+        // so we don't end up with /api/v3/v1/chat/completions (404).
+        if (baseUrl != null && BASE_URL_VERSION_SUFFIX.matcher(baseUrl).matches() && path.startsWith("/v1/")) {
             path = path.substring(3);
         }
         return path;
     }
 
+    /**
+     * Resolve the OpenAI-compatible {@code /v1/models} path against a base URL,
+     * stripping the {@code /v1} prefix when the base already carries a {@code /v{N}}
+     * suffix (Volcano Engine Ark, etc.).
+     */
+    private String resolveModelsPath(String baseUrl) {
+        String path = "/v1/models";
+        if (baseUrl != null && BASE_URL_VERSION_SUFFIX.matcher(baseUrl).matches()) {
+            path = "/models";
+        }
+        return path;
+    }
+
     private String normalizeBaseUrl(String baseUrl) {
         if (!StringUtils.hasText(baseUrl)) {
             return null;
diff --git a/mateclaw-server/src/main/resources/db/data-en.sql b/mateclaw-server/src/main/resources/db/data-en.sql
index 6046b6c6..b7dcf7a2 100644
--- a/mateclaw-server/src/main/resources/db/data-en.sql
+++ b/mateclaw-server/src/main/resources/db/data-en.sql
@@ -112,7 +112,7 @@ VALUES ('zhipu-intl', 'Zhipu AI (International)', '', 'OpenAIChatModel', '', 'ht
 
 MERGE INTO mate_model_provider (provider_id, name, api_key_prefix, chat_model, api_key, base_url, generate_kwargs, is_custom, is_local, support_model_discovery, support_connection_check, freeze_url, require_api_key, create_time, update_time)
 KEY (provider_id)
-VALUES ('volcengine', 'Volcano Engine', '', 'OpenAIChatModel', '', 'https://ark.cn-beijing.volces.com/api/v3', '{}', FALSE, FALSE, FALSE, TRUE, TRUE, TRUE, NOW(), NOW());
+VALUES ('volcengine', 'Volcano Engine', '', 'OpenAIChatModel', '', 'https://ark.cn-beijing.volces.com/api/v3', '{}', FALSE, FALSE, TRUE, TRUE, TRUE, TRUE, NOW(), NOW());
 
 MERGE INTO mate_model_provider (provider_id, name, api_key_prefix, chat_model, api_key, base_url, generate_kwargs, is_custom, is_local, support_model_discovery, support_connection_check, freeze_url, require_api_key, auth_type, create_time, update_time)
 KEY (provider_id)
@@ -248,12 +248,11 @@ MERGE INTO mate_model_config (id, name, provider, model_name, description, tempe
 (1000000221, 'GLM-5V-Turbo', 'zhipu-intl', 'glm-5v-turbo', 'Multimodal vision model (International, recommended)', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
 (1000000222, 'GLM-5', 'zhipu-intl', 'glm-5', 'Flagship model (International)', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
 (1000000223, 'GLM-5.1', 'zhipu-intl', 'glm-5.1', 'Latest flagship model (International)', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
-(1000000230, 'Doubao 1.5 Pro 256K', 'volcengine', 'doubao-1.5-pro-256k', 'Doubao flagship model with 256K context', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
-(1000000231, 'Doubao 1.5 Pro 32K', 'volcengine', 'doubao-1.5-pro-32k', 'Doubao flagship model with 32K context', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
-(1000000232, 'Doubao 1.5 Lite 32K', 'volcengine', 'doubao-1.5-lite-32k', 'Doubao lite model, cost-effective', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
-(1000000233, 'Doubao 1.5 Vision Pro 32K', 'volcengine', 'doubao-1.5-vision-pro-32k', 'Doubao multimodal vision model', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
-(1000000234, 'Doubao 1.5 Thinking Pro', 'volcengine', 'doubao-1.5-thinking-pro', 'Doubao deep reasoning model', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
-(1000000235, 'Doubao 1.5 Thinking Lite', 'volcengine', 'doubao-1.5-thinking-lite', 'Doubao lite reasoning model', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
+(1000000310, 'Doubao Seed 1.8', 'volcengine', 'doubao-seed-1-8-251228', 'Doubao flagship multimodal model, text + image, 256K context', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
+(1000000311, 'Doubao Seed Code Preview', 'volcengine', 'doubao-seed-code-preview-251028', 'Doubao code preview model, text + image, 256K context', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
+(1000000312, 'Kimi K2.5', 'volcengine', 'kimi-k2-5-260127', 'Kimi K2.5 (hosted on Volcano Ark), text + image, 256K context', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
+(1000000313, 'GLM 4.7', 'volcengine', 'glm-4-7-251222', 'GLM 4.7 (hosted on Volcano Ark), text + image, 200K context', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
+(1000000314, 'DeepSeek V3.2', 'volcengine', 'deepseek-v3-2-251201', 'DeepSeek V3.2 (hosted on Volcano Ark), text + image, 128K context', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
 (1000000240, 'Kimi for Coding', 'kimi-code', 'kimi-for-coding', 'Kimi Code dedicated coding model', 0.2, 32768, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
 (1000000250, 'GPT-5.4', 'openai-chatgpt', 'gpt-5.4', 'ChatGPT Plus/Pro member model (OAuth login)', NULL, 128000, NULL, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
 (1000000251, 'GPT-5.4 Mini', 'openai-chatgpt', 'gpt-5.4-mini', 'ChatGPT member lightweight model', NULL, 128000, NULL, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
diff --git a/mateclaw-server/src/main/resources/db/data-mysql-en.sql b/mateclaw-server/src/main/resources/db/data-mysql-en.sql
index eb14204d..0df4a308 100644
--- a/mateclaw-server/src/main/resources/db/data-mysql-en.sql
+++ b/mateclaw-server/src/main/resources/db/data-mysql-en.sql
@@ -127,7 +127,7 @@ VALUES ('zhipu-intl', 'Zhipu AI (International)', '', 'OpenAIChatModel', '', 'ht
 ON DUPLICATE KEY UPDATE name=VALUES(name), api_key_prefix=VALUES(api_key_prefix), chat_model=VALUES(chat_model), api_key=VALUES(api_key), base_url=VALUES(base_url), generate_kwargs=VALUES(generate_kwargs), is_custom=VALUES(is_custom), is_local=VALUES(is_local), support_model_discovery=VALUES(support_model_discovery), support_connection_check=VALUES(support_connection_check), freeze_url=VALUES(freeze_url), require_api_key=VALUES(require_api_key), update_time=VALUES(update_time);
 
 INSERT INTO mate_model_provider (provider_id, name, api_key_prefix, chat_model, api_key, base_url, generate_kwargs, is_custom, is_local, support_model_discovery, support_connection_check, freeze_url, require_api_key, create_time, update_time)
-VALUES ('volcengine', 'Volcano Engine', '', 'OpenAIChatModel', '', 'https://ark.cn-beijing.volces.com/api/v3', '{}', FALSE, FALSE, FALSE, TRUE, TRUE, TRUE, NOW(), NOW())
+VALUES ('volcengine', 'Volcano Engine', '', 'OpenAIChatModel', '', 'https://ark.cn-beijing.volces.com/api/v3', '{}', FALSE, FALSE, TRUE, TRUE, TRUE, TRUE, NOW(), NOW())
 ON DUPLICATE KEY UPDATE name=VALUES(name), api_key_prefix=VALUES(api_key_prefix), chat_model=VALUES(chat_model), api_key=VALUES(api_key), base_url=VALUES(base_url), generate_kwargs=VALUES(generate_kwargs), is_custom=VALUES(is_custom), is_local=VALUES(is_local), support_model_discovery=VALUES(support_model_discovery), support_connection_check=VALUES(support_connection_check), freeze_url=VALUES(freeze_url), require_api_key=VALUES(require_api_key), update_time=VALUES(update_time);
 
 INSERT INTO mate_model_provider (provider_id, name, api_key_prefix, chat_model, api_key, base_url, generate_kwargs, is_custom, is_local, support_model_discovery, support_connection_check, freeze_url, require_api_key, auth_type, create_time, update_time)
@@ -286,12 +286,11 @@ VALUES
 (1000000221, 'GLM-5V-Turbo', 'zhipu-intl', 'glm-5v-turbo', 'Multimodal vision model (International, recommended)', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
 (1000000222, 'GLM-5', 'zhipu-intl', 'glm-5', 'Flagship model (International)', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
 (1000000223, 'GLM-5.1', 'zhipu-intl', 'glm-5.1', 'Latest flagship model (International)', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
-(1000000230, 'Doubao 1.5 Pro 256K', 'volcengine', 'doubao-1.5-pro-256k', 'Doubao flagship model with 256K context', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
-(1000000231, 'Doubao 1.5 Pro 32K', 'volcengine', 'doubao-1.5-pro-32k', 'Doubao flagship model with 32K context', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
-(1000000232, 'Doubao 1.5 Lite 32K', 'volcengine', 'doubao-1.5-lite-32k', 'Doubao lite model, cost-effective', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
-(1000000233, 'Doubao 1.5 Vision Pro 32K', 'volcengine', 'doubao-1.5-vision-pro-32k', 'Doubao multimodal vision model', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
-(1000000234, 'Doubao 1.5 Thinking Pro', 'volcengine', 'doubao-1.5-thinking-pro', 'Doubao deep reasoning model', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
-(1000000235, 'Doubao 1.5 Thinking Lite', 'volcengine', 'doubao-1.5-thinking-lite', 'Doubao lite reasoning model', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
+(1000000310, 'Doubao Seed 1.8', 'volcengine', 'doubao-seed-1-8-251228', 'Doubao flagship multimodal model, text + image, 256K context', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
+(1000000311, 'Doubao Seed Code Preview', 'volcengine', 'doubao-seed-code-preview-251028', 'Doubao code preview model, text + image, 256K context', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
+(1000000312, 'Kimi K2.5', 'volcengine', 'kimi-k2-5-260127', 'Kimi K2.5 (hosted on Volcano Ark), text + image, 256K context', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
+(1000000313, 'GLM 4.7', 'volcengine', 'glm-4-7-251222', 'GLM 4.7 (hosted on Volcano Ark), text + image, 200K context', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
+(1000000314, 'DeepSeek V3.2', 'volcengine', 'deepseek-v3-2-251201', 'DeepSeek V3.2 (hosted on Volcano Ark), text + image, 128K context', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
 (1000000240, 'Kimi for Coding', 'kimi-code', 'kimi-for-coding', 'Kimi Code dedicated coding model', 0.2, 32768, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
 (1000000250, 'GPT-5.4', 'openai-chatgpt', 'gpt-5.4', 'ChatGPT Plus/Pro member model (OAuth login)', NULL, 128000, NULL, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
 (1000000251, 'GPT-5.4 Mini', 'openai-chatgpt', 'gpt-5.4-mini', 'ChatGPT member lightweight model', NULL, 128000, NULL, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
diff --git a/mateclaw-server/src/main/resources/db/data-mysql-zh.sql b/mateclaw-server/src/main/resources/db/data-mysql-zh.sql
index 9e5e7375..20186a44 100644
--- a/mateclaw-server/src/main/resources/db/data-mysql-zh.sql
+++ b/mateclaw-server/src/main/resources/db/data-mysql-zh.sql
@@ -127,7 +127,7 @@ VALUES ('zhipu-intl', 'Zhipu AI (International)', '', 'OpenAIChatModel', '', 'ht
 ON DUPLICATE KEY UPDATE name=VALUES(name), api_key_prefix=VALUES(api_key_prefix), chat_model=VALUES(chat_model), api_key=VALUES(api_key), base_url=VALUES(base_url), generate_kwargs=VALUES(generate_kwargs), is_custom=VALUES(is_custom), is_local=VALUES(is_local), support_model_discovery=VALUES(support_model_discovery), support_connection_check=VALUES(support_connection_check), freeze_url=VALUES(freeze_url), require_api_key=VALUES(require_api_key), update_time=VALUES(update_time);
 
 INSERT INTO mate_model_provider (provider_id, name, api_key_prefix, chat_model, api_key, base_url, generate_kwargs, is_custom, is_local, support_model_discovery, support_connection_check, freeze_url, require_api_key, create_time, update_time)
-VALUES ('volcengine', 'Volcano Engine (火山引擎)', '', 'OpenAIChatModel', '', 'https://ark.cn-beijing.volces.com/api/v3', '{}', FALSE, FALSE, FALSE, TRUE, TRUE, TRUE, NOW(), NOW())
+VALUES ('volcengine', 'Volcano Engine (火山引擎)', '', 'OpenAIChatModel', '', 'https://ark.cn-beijing.volces.com/api/v3', '{}', FALSE, FALSE, TRUE, TRUE, TRUE, TRUE, NOW(), NOW())
 ON DUPLICATE KEY UPDATE name=VALUES(name), api_key_prefix=VALUES(api_key_prefix), chat_model=VALUES(chat_model), api_key=VALUES(api_key), base_url=VALUES(base_url), generate_kwargs=VALUES(generate_kwargs), is_custom=VALUES(is_custom), is_local=VALUES(is_local), support_model_discovery=VALUES(support_model_discovery), support_connection_check=VALUES(support_connection_check), freeze_url=VALUES(freeze_url), require_api_key=VALUES(require_api_key), update_time=VALUES(update_time);
 
 INSERT INTO mate_model_provider (provider_id, name, api_key_prefix, chat_model, api_key, base_url, generate_kwargs, is_custom, is_local, support_model_discovery, support_connection_check, freeze_url, require_api_key, auth_type, create_time, update_time)
@@ -286,12 +286,11 @@ VALUES
 (1000000221, 'GLM-5V-Turbo', 'zhipu-intl', 'glm-5v-turbo', '多模态视觉模型（国际版，推荐）', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
 (1000000222, 'GLM-5', 'zhipu-intl', 'glm-5', '旗舰模型（国际版）', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
 (1000000223, 'GLM-5.1', 'zhipu-intl', 'glm-5.1', '最新旗舰模型（国际版）', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
-(1000000230, 'Doubao 1.5 Pro 256K', 'volcengine', 'doubao-1.5-pro-256k', '豆包旗舰模型，256K 超长上下文', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
-(1000000231, 'Doubao 1.5 Pro 32K', 'volcengine', 'doubao-1.5-pro-32k', '豆包旗舰模型，32K 上下文', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
-(1000000232, 'Doubao 1.5 Lite 32K', 'volcengine', 'doubao-1.5-lite-32k', '豆包轻量模型，高性价比', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
-(1000000233, 'Doubao 1.5 Vision Pro 32K', 'volcengine', 'doubao-1.5-vision-pro-32k', '豆包多模态视觉模型', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
-(1000000234, 'Doubao 1.5 Thinking Pro', 'volcengine', 'doubao-1.5-thinking-pro', '豆包深度推理模型', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
-(1000000235, 'Doubao 1.5 Thinking Lite', 'volcengine', 'doubao-1.5-thinking-lite', '豆包轻量推理模型', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
+(1000000310, 'Doubao Seed 1.8', 'volcengine', 'doubao-seed-1-8-251228', '豆包旗舰多模态模型，文本+图像，256K 上下文', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
+(1000000311, 'Doubao Seed Code Preview', 'volcengine', 'doubao-seed-code-preview-251028', '豆包代码预览模型，文本+图像，256K 上下文', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
+(1000000312, 'Kimi K2.5', 'volcengine', 'kimi-k2-5-260127', 'Kimi K2.5（火山方舟托管），文本+图像，256K 上下文', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
+(1000000313, 'GLM 4.7', 'volcengine', 'glm-4-7-251222', 'GLM 4.7（火山方舟托管），文本+图像，200K 上下文', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
+(1000000314, 'DeepSeek V3.2', 'volcengine', 'deepseek-v3-2-251201', 'DeepSeek V3.2（火山方舟托管），文本+图像，128K 上下文', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
 (1000000240, 'Kimi for Coding', 'kimi-code', 'kimi-for-coding', 'Kimi Code 专用编码模型', 0.2, 32768, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
 (1000000250, 'GPT-5.4', 'openai-chatgpt', 'gpt-5.4', 'ChatGPT Plus/Pro 会员模型（OAuth 登录）', NULL, 128000, NULL, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
 (1000000251, 'GPT-5.4 Mini', 'openai-chatgpt', 'gpt-5.4-mini', 'ChatGPT 会员轻量模型', NULL, 128000, NULL, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
diff --git a/mateclaw-server/src/main/resources/db/data-zh.sql b/mateclaw-server/src/main/resources/db/data-zh.sql
index e0e2bf9f..3431ce07 100644
--- a/mateclaw-server/src/main/resources/db/data-zh.sql
+++ b/mateclaw-server/src/main/resources/db/data-zh.sql
@@ -112,7 +112,7 @@ VALUES ('zhipu-intl', 'Zhipu AI (International)', '', 'OpenAIChatModel', '', 'ht
 
 MERGE INTO mate_model_provider (provider_id, name, api_key_prefix, chat_model, api_key, base_url, generate_kwargs, is_custom, is_local, support_model_discovery, support_connection_check, freeze_url, require_api_key, create_time, update_time)
 KEY (provider_id)
-VALUES ('volcengine', 'Volcano Engine (火山引擎)', '', 'OpenAIChatModel', '', 'https://ark.cn-beijing.volces.com/api/v3', '{}', FALSE, FALSE, FALSE, TRUE, TRUE, TRUE, NOW(), NOW());
+VALUES ('volcengine', 'Volcano Engine (火山引擎)', '', 'OpenAIChatModel', '', 'https://ark.cn-beijing.volces.com/api/v3', '{}', FALSE, FALSE, TRUE, TRUE, TRUE, TRUE, NOW(), NOW());
 
 MERGE INTO mate_model_provider (provider_id, name, api_key_prefix, chat_model, api_key, base_url, generate_kwargs, is_custom, is_local, support_model_discovery, support_connection_check, freeze_url, require_api_key, auth_type, create_time, update_time)
 KEY (provider_id)
@@ -252,12 +252,11 @@ MERGE INTO mate_model_config (id, name, provider, model_name, description, tempe
 (1000000221, 'GLM-5V-Turbo', 'zhipu-intl', 'glm-5v-turbo', '多模态视觉模型（国际版，推荐）', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
 (1000000222, 'GLM-5', 'zhipu-intl', 'glm-5', '旗舰模型（国际版）', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
 (1000000223, 'GLM-5.1', 'zhipu-intl', 'glm-5.1', '最新旗舰模型（国际版）', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
-(1000000230, 'Doubao 1.5 Pro 256K', 'volcengine', 'doubao-1.5-pro-256k', '豆包旗舰模型，256K 超长上下文', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
-(1000000231, 'Doubao 1.5 Pro 32K', 'volcengine', 'doubao-1.5-pro-32k', '豆包旗舰模型，32K 上下文', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
-(1000000232, 'Doubao 1.5 Lite 32K', 'volcengine', 'doubao-1.5-lite-32k', '豆包轻量模型，高性价比', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
-(1000000233, 'Doubao 1.5 Vision Pro 32K', 'volcengine', 'doubao-1.5-vision-pro-32k', '豆包多模态视觉模型', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
-(1000000234, 'Doubao 1.5 Thinking Pro', 'volcengine', 'doubao-1.5-thinking-pro', '豆包深度推理模型', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
-(1000000235, 'Doubao 1.5 Thinking Lite', 'volcengine', 'doubao-1.5-thinking-lite', '豆包轻量推理模型', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
+(1000000310, 'Doubao Seed 1.8', 'volcengine', 'doubao-seed-1-8-251228', '豆包旗舰多模态模型，文本+图像，256K 上下文', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
+(1000000311, 'Doubao Seed Code Preview', 'volcengine', 'doubao-seed-code-preview-251028', '豆包代码预览模型，文本+图像，256K 上下文', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
+(1000000312, 'Kimi K2.5', 'volcengine', 'kimi-k2-5-260127', 'Kimi K2.5（火山方舟托管），文本+图像，256K 上下文', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
+(1000000313, 'GLM 4.7', 'volcengine', 'glm-4-7-251222', 'GLM 4.7（火山方舟托管），文本+图像，200K 上下文', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
+(1000000314, 'DeepSeek V3.2', 'volcengine', 'deepseek-v3-2-251201', 'DeepSeek V3.2（火山方舟托管），文本+图像，128K 上下文', 0.7, 4096, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
 (1000000240, 'Kimi for Coding', 'kimi-code', 'kimi-for-coding', 'Kimi Code 专用编码模型', 0.2, 32768, 0.8, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
 (1000000250, 'GPT-5.4', 'openai-chatgpt', 'gpt-5.4', 'ChatGPT Plus/Pro 会员模型（OAuth 登录）', NULL, 128000, NULL, TRUE, TRUE, FALSE, NOW(), NOW(), 0),
 (1000000251, 'GPT-5.4 Mini', 'openai-chatgpt', 'gpt-5.4-mini', 'ChatGPT 会员轻量模型', NULL, 128000, NULL, TRUE, TRUE, FALSE, NOW(), NOW(), 0),