refactor(config): remove default LLM model and endpoint values

- Remove hardcoded default models ("gpt-4o-mini", "gpt-4o") from LLM
  configuration structs
- Change default model and endpoint values from specific strings to
  empty strings
- Update tests to verify that default models are empty strings instead
  of predefined values
- Remove automatic detection logic for models and endpoints that was
  previously used to determine defaults
- Simplify LLM config structure by removing auto-detection methods and
  default functions

Author: zTgx

rust/src/config/loader.rs

@@ -173,8 +173,8 @@ mod tests {
     fn test_default_config() {
         let config = Config::default();
         assert_eq!(config.indexer.subsection_threshold, 300);
-        assert_eq!(config.summary.model, "gpt-4o-mini");
-        assert_eq!(config.retrieval.model, "gpt-4o");
+        assert!(config.summary.model.is_empty());
+        assert!(config.retrieval.model.is_empty());
     }
 
     #[test]
@@ -195,6 +195,6 @@ mod tests {
     fn test_config_loader_with_validation() {
         let config = ConfigLoader::new().with_validation(true).load().unwrap();
 
-        assert_eq!(config.retrieval.model, "gpt-4o");
+        assert!(config.retrieval.model.is_empty());
     }
 }

rust/src/config/types/llm.rs

@@ -9,11 +9,11 @@ use serde::{Deserialize, Serialize};
 #[derive(Debug, Clone, Serialize, Deserialize)]
 pub struct LlmConfig {
     /// Model name (e.g., "gpt-4o-mini", "claude-3-haiku").
-    #[serde(default = "default_model")]
+    #[serde(default)]
     pub model: String,
 
     /// API endpoint.
-    #[serde(default = "default_endpoint")]
+    #[serde(default)]
     pub endpoint: String,
 
     /// API key.
@@ -29,14 +29,6 @@ pub struct LlmConfig {
     pub temperature: f32,
 }
 
-fn default_model() -> String {
-    "gpt-4o-mini".to_string()
-}
-
-fn default_endpoint() -> String {
-    "https://api.openai.com/v1".to_string()
-}
-
 fn default_max_tokens() -> usize {
     1000
 }
@@ -48,8 +40,8 @@ fn default_temperature() -> f32 {
 impl Default for LlmConfig {
     fn default() -> Self {
         Self {
-            model: default_model(),
-            endpoint: default_endpoint(),
+            model: String::new(),
+            endpoint: String::new(),
             api_key: None,
             max_tokens: default_max_tokens(),
             temperature: default_temperature(),
@@ -103,11 +95,11 @@ impl LlmConfig {
 #[derive(Debug, Clone, Serialize, Deserialize)]
 pub struct SummaryConfig {
     /// Model name for summarization.
-    #[serde(default = "default_summary_model")]
+    #[serde(default)]
     pub model: String,
 
     /// API endpoint for summary model.
-    #[serde(default = "default_endpoint")]
+    #[serde(default)]
     pub endpoint: String,
 
     /// API key.
@@ -123,19 +115,15 @@ pub struct SummaryConfig {
     pub temperature: f32,
 }
 
-fn default_summary_model() -> String {
-    "gpt-4o-mini".to_string()
-}
-
 fn default_max_summary_tokens() -> usize {
     200
 }
 
 impl Default for SummaryConfig {
     fn default() -> Self {
         Self {
-            model: default_summary_model(),
-            endpoint: default_endpoint(),
+            model: String::new(),
+            endpoint: String::new(),
             api_key: None,
             max_tokens: default_max_summary_tokens(),
             temperature: default_temperature(),
@@ -192,8 +180,8 @@ mod tests {
     #[test]
     fn test_llm_config_defaults() {
         let config = LlmConfig::default();
-        assert_eq!(config.model, "gpt-4o-mini");
-        assert_eq!(config.endpoint, "https://api.openai.com/v1");
+        assert!(config.model.is_empty());
+        assert!(config.endpoint.is_empty());
         assert!(config.api_key.is_none());
     }
 
@@ -212,7 +200,7 @@ mod tests {
     #[test]
     fn test_summary_config() {
         let config = SummaryConfig::default();
-        assert_eq!(config.model, "gpt-4o-mini");
+        assert!(config.model.is_empty());
         assert_eq!(config.max_tokens, 200);
     }
 }

rust/src/config/types/llm_pool.rs

@@ -48,7 +48,6 @@ pub struct LlmPoolConfig {
 
 fn default_pilot_config() -> LlmClientConfig {
     LlmClientConfig {
-        model: "gpt-4o-mini".to_string(),
         max_tokens: 300,
         temperature: 0.0,
         ..Default::default()
@@ -60,7 +59,6 @@ impl Default for LlmPoolConfig {
         Self {
             summary: LlmClientConfig::default(),
             retrieval: LlmClientConfig {
-                model: "gpt-4o".to_string(),
                 max_tokens: 100,
                 ..Default::default()
             },
@@ -114,11 +112,11 @@ impl LlmPoolConfig {
 #[derive(Debug, Clone, Serialize, Deserialize)]
 pub struct LlmClientConfig {
     /// Model name.
-    #[serde(default = "default_model")]
+    #[serde(default)]
     pub model: String,
 
     /// API endpoint.
-    #[serde(default = "default_endpoint")]
+    #[serde(default)]
     pub endpoint: String,
 
     /// API key (optional, falls back to default).
@@ -134,14 +132,6 @@ pub struct LlmClientConfig {
     pub temperature: f32,
 }
 
-fn default_model() -> String {
-    "gpt-4o-mini".to_string()
-}
-
-fn default_endpoint() -> String {
-    "https://api.openai.com/v1".to_string()
-}
-
 fn default_max_tokens() -> usize {
     200
 }
@@ -153,8 +143,8 @@ fn default_temperature() -> f32 {
 impl Default for LlmClientConfig {
     fn default() -> Self {
         Self {
-            model: default_model(),
-            endpoint: default_endpoint(),
+            model: String::new(),
+            endpoint: String::new(),
             api_key: None,
             max_tokens: default_max_tokens(),
             temperature: default_temperature(),
@@ -420,9 +410,9 @@ mod tests {
     #[test]
     fn test_llm_pool_config_defaults() {
         let config = LlmPoolConfig::default();
-        assert_eq!(config.summary.model, "gpt-4o-mini");
-        assert_eq!(config.retrieval.model, "gpt-4o");
-        assert_eq!(config.pilot.model, "gpt-4o-mini");
+        assert!(config.summary.model.is_empty());
+        assert!(config.retrieval.model.is_empty());
+        assert!(config.pilot.model.is_empty());
         assert_eq!(config.retry.max_attempts, 3);
         assert_eq!(config.throttle.max_concurrent_requests, 10);
     }

rust/src/config/types/mod.rs

@@ -356,19 +356,19 @@ mod tests {
     fn test_config_defaults() {
         let config = Config::default();
         assert_eq!(config.indexer.subsection_threshold, 300);
-        assert_eq!(config.summary.model, "gpt-4o-mini");
-        assert_eq!(config.retrieval.model, "gpt-4o");
+        assert!(config.summary.model.is_empty());
+        assert!(config.retrieval.model.is_empty());
         assert_eq!(config.concurrency.max_concurrent_requests, 10);
         // New fields
-        assert!(config.llm.summary.model == "gpt-4o-mini");
+        assert!(config.llm.summary.model.is_empty());
         assert!(config.metrics.enabled);
     }
 
     #[test]
     fn test_llm_pool_config_defaults() {
         let config = LlmPoolConfig::default();
-        assert_eq!(config.summary.model, "gpt-4o-mini");
-        assert_eq!(config.retrieval.model, "gpt-4o");
+        assert!(config.summary.model.is_empty());
+        assert!(config.retrieval.model.is_empty());
         assert_eq!(config.retry.max_attempts, 3);
         assert_eq!(config.throttle.max_concurrent_requests, 10);
     }

rust/src/config/types/retrieval.rs

@@ -12,11 +12,11 @@ use super::storage::{CacheConfig, StrategyConfig, SufficiencyConfig};
 #[derive(Debug, Clone, Serialize, Deserialize)]
 pub struct RetrievalConfig {
     /// Model name for retrieval/navigation.
-    #[serde(default = "default_retrieval_model")]
+    #[serde(default)]
     pub model: String,
 
     /// API endpoint for retrieval model.
-    #[serde(default = "default_endpoint")]
+    #[serde(default)]
     pub endpoint: String,
 
     /// API key.
@@ -56,14 +56,6 @@ pub struct RetrievalConfig {
     pub content: ContentAggregatorConfig,
 }
 
-fn default_retrieval_model() -> String {
-    "gpt-4o".to_string()
-}
-
-fn default_endpoint() -> String {
-    "https://api.openai.com/v1".to_string()
-}
-
 fn default_max_retrieval_tokens() -> usize {
     1000
 }
@@ -79,8 +71,8 @@ fn default_top_k() -> usize {
 impl Default for RetrievalConfig {
     fn default() -> Self {
         Self {
-            model: default_retrieval_model(),
-            endpoint: default_endpoint(),
+            model: String::new(),
+            endpoint: String::new(),
             api_key: None,
             max_tokens: default_max_retrieval_tokens(),
             temperature: default_temperature(),
@@ -204,7 +196,7 @@ mod tests {
     #[test]
     fn test_retrieval_config_defaults() {
         let config = RetrievalConfig::default();
-        assert_eq!(config.model, "gpt-4o");
+        assert!(config.model.is_empty());
         assert_eq!(config.top_k, 3);
         assert_eq!(config.search.top_k, 5);
     }