fix: Add better error handling and logging to BayerMGA provider

- Added comprehensive logging for model fetching and instance creation
- Improved error handling with detailed error messages
- Added input validation for API responses
- Better debugging for inference endpoint issues

app/lib/modules/llm/providers/bayer-mga.ts

@@ -2,6 +2,9 @@ import { BaseProvider, getOpenAILikeModel } from '~/lib/modules/llm/base-provide
 import type { ModelInfo } from '~/lib/modules/llm/types';
 import type { IProviderSetting } from '~/types/model';
 import type { LanguageModelV1 } from 'ai';
+import { createScopedLogger } from '~/utils/logger';
+
+const logger = createScopedLogger('BayerMGAProvider');
 
 export default class BayerMGAProvider extends BaseProvider {
   name = 'BayerMGA';
@@ -20,35 +23,58 @@ export default class BayerMGAProvider extends BaseProvider {
     settings?: IProviderSetting,
     serverEnv: Record<string, string> = {},
   ): Promise<ModelInfo[]> {
-    const { baseUrl, apiKey } = this.getProviderBaseUrlAndKey({
-      apiKeys,
-      providerSettings: settings,
-      serverEnv,
-      defaultBaseUrlKey: 'BAYER_MGA_API_BASE_URL',
-      defaultApiTokenKey: 'BAYER_MGA_API_KEY',
-    });
+    try {
+      const { baseUrl, apiKey } = this.getProviderBaseUrlAndKey({
+        apiKeys,
+        providerSettings: settings,
+        serverEnv,
+        defaultBaseUrlKey: 'BAYER_MGA_API_BASE_URL',
+        defaultApiTokenKey: 'BAYER_MGA_API_KEY',
+      });
 
-    if (!baseUrl || !apiKey) {
+      if (!baseUrl || !apiKey) {
+        logger.warn('Missing baseUrl or apiKey configuration for Bayer MGA provider');
+        return [];
+      }
+
+      const modelsUrl = `${baseUrl}/models?include_hidden_models=false&include_aliases=true`;
+      logger.info(`Fetching models from ${modelsUrl}`);
+
+      const response = await fetch(modelsUrl, {
+        headers: {
+          Authorization: `Bearer ${apiKey}`,
+        },
+      });
+
+      if (!response.ok) {
+        const errorText = await response.text();
+        logger.error(`Failed to fetch Bayer MGA models: ${response.status} ${errorText}`);
+        return [];
+      }
+
+      const res = (await response.json()) as any;
+
+      if (!res.data || !Array.isArray(res.data)) {
+        logger.error(`Invalid response format from Bayer MGA API: missing data array`);
+        return [];
+      }
+
+      // Filter for available models and map to ModelInfo format
+      const models = res.data
+        .filter((model: any) => model.model_status === 'available')
+        .map((model: any) => ({
+          name: model.model,
+          label: model.name || model.model,
+          provider: this.name,
+          maxTokenAllowed: model.context_window || 8000,
+        }));
+
+      logger.info(`Found ${models.length} available models from Bayer MGA`);
+      return models;
+    } catch (error) {
+      logger.error(`Error fetching Bayer MGA models: ${error instanceof Error ? error.message : String(error)}`);
       return [];
     }
-
-    const response = await fetch(`${baseUrl}/models?include_hidden_models=false&include_aliases=true`, {
-      headers: {
-        Authorization: `Bearer ${apiKey}`,
-      },
-    });
-
-    const res = (await response.json()) as any;
-
-    // Filter for available models and map to ModelInfo format
-    return res.data
-      .filter((model: any) => model.model_status === 'available')
-      .map((model: any) => ({
-        name: model.model,
-        label: model.name || model.model,
-        provider: this.name,
-        maxTokenAllowed: model.context_window || 8000,
-      }));
   }
 
   getModelInstance(options: {
@@ -57,20 +83,31 @@ export default class BayerMGAProvider extends BaseProvider {
     apiKeys?: Record<string, string>;
     providerSettings?: Record<string, IProviderSetting>;
   }): LanguageModelV1 {
-    const { model, serverEnv, apiKeys, providerSettings } = options;
+    try {
+      const { model, serverEnv, apiKeys, providerSettings } = options;
 
-    const { baseUrl, apiKey } = this.getProviderBaseUrlAndKey({
-      apiKeys,
-      providerSettings: providerSettings?.[this.name],
-      serverEnv: serverEnv as any,
-      defaultBaseUrlKey: 'BAYER_MGA_API_BASE_URL',
-      defaultApiTokenKey: 'BAYER_MGA_API_KEY',
-    });
+      const { baseUrl, apiKey } = this.getProviderBaseUrlAndKey({
+        apiKeys,
+        providerSettings: providerSettings?.[this.name],
+        serverEnv: serverEnv as any,
+        defaultBaseUrlKey: 'BAYER_MGA_API_BASE_URL',
+        defaultApiTokenKey: 'BAYER_MGA_API_KEY',
+      });
 
-    if (!baseUrl || !apiKey) {
-      throw new Error(`Missing configuration for ${this.name} provider`);
+      if (!baseUrl || !apiKey) {
+        throw new Error(`Missing configuration for ${this.name} provider`);
+      }
+
+      // Ensure baseUrl doesn't have trailing slash since the SDK will append paths
+      const normalizedBaseUrl = baseUrl.endsWith('/') ? baseUrl.slice(0, -1) : baseUrl;
+      
+      logger.info(`Creating model instance for ${model} using Bayer MGA API at ${normalizedBaseUrl}`);
+
+      // The OpenAI SDK will automatically append /chat/completions to the baseURL
+      return getOpenAILikeModel(normalizedBaseUrl, apiKey, model);
+    } catch (error) {
+      logger.error(`Error creating Bayer MGA model instance: ${error instanceof Error ? error.message : String(error)}`);
+      throw error;
     }
-
-    return getOpenAILikeModel(baseUrl, apiKey, model);
   }
 }