configure using different models for chat

go-backend/handlers/chat.go

@@ -151,16 +151,23 @@ func (s *Handler) PostChatMessageRoute(w http.ResponseWriter, r *http.Request) {
 		http.Error(w, err.Error(), http.StatusInternalServerError)
 		return
 	}
-	message, err = s.GetChatCompletion(userID, message.ConversationID)
+
+	model, err := s.QueryLLMModel(message.ConfigurationID)
+	if err != nil {
+		log.Printf("error getting model: %v", err)
+		http.Error(w, err.Error(), http.StatusInternalServerError)
+		return
+	}
+	client := llms.NewClientFromModel(s.DB, model)
+
+	message, err = s.GetChatCompletion(userID, client, message.ConversationID)
 	if err != nil {
 		log.Printf("error getting chat completion: %v", err)
 		http.Error(w, err.Error(), http.StatusInternalServerError)
 		return
 	}
 
 	if newConversation {
-		client := llms.NewDefaultClient(s.DB)
-
 		summary, err := llms.CreateConversationSummary(client, message)
 		if err != nil {
 			http.Error(w, err.Error(), http.StatusInternalServerError)
@@ -206,8 +213,54 @@ func (s *Handler) WriteConversationSummary(userID int, summary models.Conversati
 	return nil
 }
 
+func (s *Handler) QueryLLMModel(configurationID int) (models.LLMModel, error) {
+	log.Printf("querying llm model: %v", configurationID)
+	query := `
+        SELECT 
+            m.id, 
+            m.name, 
+            m.model_identifier, 
+            m.description, 
+            m.is_active,
+            p.id as provider_id,
+            p.name as provider_name,
+            p.base_url,
+            p.api_key_required,
+            p.api_key
+        FROM llm_models m
+        JOIN user_llm_configurations uc ON m.id = uc.model_id
+        JOIN llm_providers p ON m.provider_id = p.id
+        WHERE uc.id = $1
+    `
+	var model models.LLMModel
+	var provider models.LLMProvider
+
+	err := s.DB.QueryRow(query, configurationID).Scan(
+		&model.ID,
+		&model.Name,
+		&model.ModelIdentifier,
+		&model.Description,
+		&model.IsActive,
+		&provider.ID,
+		&provider.Name,
+		&provider.BaseURL,
+		&provider.APIKeyRequired,
+		&provider.APIKey,
+	)
+	if err != nil {
+		log.Printf("error querying llm model: %v", err)
+		return models.LLMModel{}, err
+	}
+
+	model.Provider = &provider
+	model.ProviderID = provider.ID
+
+	return model, nil
+}
+
 func (s *Handler) AddChatMessage(userID int, message models.ChatCompletion) (models.ChatCompletion, error) {
 	var nextSequence int
+
 	err := s.DB.QueryRow(`
         SELECT COALESCE(MAX(sequence_number), 0) + 1
         FROM chat_completions
@@ -260,7 +313,7 @@ func (s *Handler) AddChatMessage(userID int, message models.ChatCompletion) (mod
 	return insertedMessage, nil
 }
 
-func (s *Handler) GetChatCompletion(userID int, conversationID string) (models.ChatCompletion, error) {
+func (s *Handler) GetChatCompletion(userID int, client *models.LLMClient, conversationID string) (models.ChatCompletion, error) {
 
 	messages, err := s.GetChatMessagesInConversation(userID, conversationID)
 	if err != nil {
@@ -280,8 +333,6 @@ func (s *Handler) GetChatCompletion(userID int, conversationID string) (models.C
 		return models.ChatCompletion{}, fmt.Errorf("failed to process response")
 	}
 
-	client := llms.NewDefaultClient(s.DB)
-
 	completion, err := llms.ChatCompletion(client, messages)
 	if err != nil {
 		log.Printf("error generating chat completion: %v", err)

go-backend/llms/chat.go

@@ -15,7 +15,7 @@ func ChatCompletion(c *models.LLMClient, pastMessages []models.ChatCompletion) (
 		return models.ChatCompletion{
 			Role:    "assistant",
 			Content: "This is a mock response for testing",
-			Model:   models.MODEL,
+			Model:   c.Model.ModelIdentifier,
 			Tokens:  100,
 		}, nil
 	}
@@ -82,7 +82,7 @@ func CreateConversationSummary(c *models.LLMClient, message models.ChatCompletio
 		ID:        id,
 		Title:     resp.Choices[0].Message.Content,
 		CreatedAt: created,
-		Model:     models.MODEL,
+		Model:     c.Model.ModelIdentifier,
 	}
 	return result, nil
 }

go-backend/llms/client.go

@@ -10,6 +10,15 @@ import (
 	openai "github.com/sashabaranov/go-openai"
 )
 
+func NewClientFromModel(db *sql.DB, model models.LLMModel) *models.LLMClient {
+	config := openai.DefaultConfig(model.Provider.APIKey)
+	config.BaseURL = model.Provider.BaseURL
+
+	client := NewClient(db, config)
+	client.Model = &model
+	return client
+}
+
 func NewDefaultClient(db *sql.DB) *models.LLMClient {
 	config := openai.DefaultConfig(os.Getenv("ZETTEL_LLM_KEY"))
 	config.BaseURL = os.Getenv("ZETTEL_LLM_ENDPOINT")
@@ -43,7 +52,7 @@ func ExecuteLLMRequest(c *models.LLMClient, messages []openai.ChatCompletionMess
 	resp, err := c.Client.CreateChatCompletion(
 		context.Background(),
 		openai.ChatCompletionRequest{
-			Model:    models.MODEL,
+			Model:    c.Model.ModelIdentifier,
 			Messages: messages,
 		},
 	)

go-backend/models/chat.go

@@ -13,6 +13,7 @@ type LLMProvider struct {
 	Name           string    `json:"name"`
 	BaseURL        string    `json:"base_url"`
 	APIKeyRequired bool      `json:"api_key_required"`
+	APIKey         string    `json:"api_key,omitempty"`
 	CreatedAt      time.Time `json:"created_at"`
 	UpdatedAt      time.Time `json:"updated_at"`
 }
@@ -60,6 +61,7 @@ type LLMClient struct {
 	Client         *openai.Client
 	Testing        bool
 	EmbeddingQueue *LLMRequestQueue
+	Model          *LLMModel
 }
 
 func NewEmbeddingQueue(db *sql.DB) *LLMRequestQueue {
@@ -115,7 +117,7 @@ type ChatCompletion struct {
 	ReferencedCardPKs []int         `json:"referenced_card_pks"`
 	ReferencedCards   []PartialCard `json:"cards"`
 	UserQuery         string        `json:"user_query"`
-	ModelID           int           `json:"model_id"`
+	ConfigurationID   int           `json:"configuration_id"`
 }
 
 type ChatData struct {

zettelkasten-front/src/api/chat.ts

@@ -42,7 +42,7 @@ export function postChatMessage(
     conversation_id: conversationId, // Will be undefined for new conversations
     user_query: content,
     referenced_card_pks: contextCards?.map((card) => card.id),
-    model_id: configurationId, 
+    configuration_id: configurationId, 
   };
 
   return fetch(url, {