diff --git a/RequestFirework.go b/RequestFirework.go
index 49d74b9..a1ce18c 100644
--- a/RequestFirework.go
+++ b/RequestFirework.go
@@ -14,6 +14,7 @@ import (
 type FireworkChatCompletionRequest struct {
 	Model       string           `json:"model"`
 	Messages    []RequestMessage `json:"messages"`
+  MaxTokens   int              `json:"max_tokens"`
 	Temperature float64          `json:"temperature"`
 }
 
@@ -41,7 +42,7 @@ type FireworkChoice struct {
 func addFireworkMessage(c *fiber.Ctx, llm LLM, selected bool) edgedb.UUID {
 	Messages := getAllSelectedMessages(c)
 
-	chatCompletion, err := RequestFirework(c, llm.Model.ModelID, Messages, float64(llm.Temperature), llm.Context)
+	chatCompletion, err := RequestFirework(c, llm.Model.ModelID, Messages, float64(llm.Temperature), llm.Context, int(llm.MaxToken))
 	if err != nil {
     fmt.Println("Error requesting Firework: ", err)
     id := insertBotMessage(c, "Error requesting Firework, model may not be available anymore. Better error message in development.", selected, llm.ID)
@@ -110,7 +111,7 @@ func TestFireworkKey(apiKey string) bool {
 	return true
 }
 
-func RequestFirework(c *fiber.Ctx, model string, messages []Message, temperature float64, context string) (FireworkChatCompletionResponse, error) {
+func RequestFirework(c *fiber.Ctx, model string, messages []Message, temperature float64, context string, maxTokens int) (FireworkChatCompletionResponse, error) {
 	var apiKey string
 	err := edgeGlobalClient.WithGlobals(map[string]interface{}{"ext::auth::client_token": c.Cookies("jade-edgedb-auth-token")}).QuerySingle(edgeCtx, `
 		with
@@ -130,6 +131,7 @@ func RequestFirework(c *fiber.Ctx, model string, messages []Message, temperature
 	requestBody := FireworkChatCompletionRequest{
 		Model:       "accounts/fireworks/models/" + model,
 		Messages:    Message2RequestMessage(messages, context),
+    MaxTokens:   maxTokens,
 		Temperature: temperature,
 	}
 
diff --git a/RequestGroq.go b/RequestGroq.go
index 62028b4..7dd3445 100644
--- a/RequestGroq.go
+++ b/RequestGroq.go
@@ -14,6 +14,7 @@ import (
 type GroqChatCompletionRequest struct {
 	Model       string           `json:"model"`
 	Messages    []RequestMessage `json:"messages"`
+  MaxTokens   int              `json:"max_tokens"`
 	Temperature float64          `json:"temperature"`
 }
 
@@ -41,7 +42,7 @@ type GroqChoice struct {
 func addGroqMessage(c *fiber.Ctx, llm LLM, selected bool) edgedb.UUID {
 	Messages := getAllSelectedMessages(c)
 
-	chatCompletion, err := RequestGroq(c, llm.Model.ModelID, Messages, float64(llm.Temperature), llm.Context)
+	chatCompletion, err := RequestGroq(c, llm.Model.ModelID, Messages, float64(llm.Temperature), llm.Context, int(llm.MaxToken))
 	if err != nil {
     fmt.Println("Error requesting Groq: ", err)
     id := insertBotMessage(c, "Error requesting Groq, model may not be available anymore. Better error message in development.", selected, llm.ID)
@@ -110,7 +111,7 @@ func TestGroqKey(apiKey string) bool {
 	return true
 }
 
-func RequestGroq(c *fiber.Ctx, model string, messages []Message, temperature float64, context string) (GroqChatCompletionResponse, error) {
+func RequestGroq(c *fiber.Ctx, model string, messages []Message, temperature float64, context string, maxTokens int) (GroqChatCompletionResponse, error) {
 	var apiKey string
 	err := edgeGlobalClient.WithGlobals(map[string]interface{}{"ext::auth::client_token": c.Cookies("jade-edgedb-auth-token")}).QuerySingle(edgeCtx, `
 		with
@@ -130,6 +131,7 @@ func RequestGroq(c *fiber.Ctx, model string, messages []Message, temperature flo
 	requestBody := GroqChatCompletionRequest{
 		Model:       model,
 		Messages:    Message2RequestMessage(messages, context),
+    MaxTokens:   maxTokens,
 		Temperature: temperature,
 	}
 
diff --git a/RequestHuggingface.go b/RequestHuggingface.go
index 3d12b10..aaf346c 100644
--- a/RequestHuggingface.go
+++ b/RequestHuggingface.go
@@ -15,6 +15,7 @@ type HuggingfaceChatCompletionRequest struct {
 	Model       string           `json:"model"`
 	Messages    []RequestMessage `json:"messages"`
 	Temperature float64          `json:"temperature"`
+  MaxTokens   int              `json:"max_tokens"`
 	Stream      bool             `json:"stream"`
 }
 
@@ -41,7 +42,7 @@ type HuggingfaceChoice struct {
 func addHuggingfaceMessage(c *fiber.Ctx, llm LLM, selected bool) edgedb.UUID {
 	Messages := getAllSelectedMessages(c)
 
-	chatCompletion, err := RequestHuggingface(c, llm, Messages, float64(llm.Temperature))
+	chatCompletion, err := RequestHuggingface(c, llm, Messages, float64(llm.Temperature), int(llm.MaxToken))
 	if err != nil {
     fmt.Println("Error requesting Huggingface: ", err)
     id := insertBotMessage(c, "Error requesting Huggingface.", selected, llm.ID)
@@ -57,13 +58,14 @@ func addHuggingfaceMessage(c *fiber.Ctx, llm LLM, selected bool) edgedb.UUID {
 	}
 }
 
-func RequestHuggingface(c *fiber.Ctx, llm LLM, messages []Message, temperature float64) (HuggingfaceChatCompletionResponse, error) {
+func RequestHuggingface(c *fiber.Ctx, llm LLM, messages []Message, temperature float64, maxTokens int) (HuggingfaceChatCompletionResponse, error) {
 	url := llm.Endpoint.Endpoint
 
 	requestBody := HuggingfaceChatCompletionRequest{
 		Model:       "tgi",
 		Messages:    Message2RequestMessage(messages, llm.Context),
 		Temperature: temperature,
+    MaxTokens:   maxTokens,
 		Stream:      false,
 	}
 
diff --git a/RequestMistral.go b/RequestMistral.go
index 2147028..ae684d8 100644
--- a/RequestMistral.go
+++ b/RequestMistral.go
@@ -14,6 +14,7 @@ import (
 type MistralChatCompletionRequest struct {
 	Model       string           `json:"model"`
 	Messages    []RequestMessage `json:"messages"`
+  MaxTokens   int              `json:"max_tokens"`
 	Temperature float64          `json:"temperature"`
 }
 type MistralChatCompletionResponse struct {
@@ -40,7 +41,7 @@ type MistralChoice struct {
 func addMistralMessage(c *fiber.Ctx, llm LLM, selected bool) edgedb.UUID {
 	Messages := getAllSelectedMessages(c)
 
-	chatCompletion, err := RequestMistral(c, llm.Model.ModelID, Messages, float64(llm.Temperature), llm.Context)
+	chatCompletion, err := RequestMistral(c, llm.Model.ModelID, Messages, float64(llm.Temperature), llm.Context, int(llm.MaxToken))
 	if err != nil {
     fmt.Println("Error requesting Mistral: ", err)
     id := insertBotMessage(c, "Error requesting Mistral, model may not be available anymore. Better error message in development.", selected, llm.ID)
@@ -115,7 +116,7 @@ func TestMistralKey(apiKey string) bool {
 	return true
 }
 
-func RequestMistral(c *fiber.Ctx, model string, messages []Message, temperature float64, context string) (MistralChatCompletionResponse, error) {
+func RequestMistral(c *fiber.Ctx, model string, messages []Message, temperature float64, context string, maxTokens int) (MistralChatCompletionResponse, error) {
 	var apiKey string
 	err := edgeGlobalClient.WithGlobals(map[string]interface{}{"ext::auth::client_token": c.Cookies("jade-edgedb-auth-token")}).QuerySingle(edgeCtx, `
 		with
@@ -135,6 +136,7 @@ func RequestMistral(c *fiber.Ctx, model string, messages []Message, temperature
 	requestBody := MistralChatCompletionRequest{
 		Model:       model,
 		Messages:    Message2RequestMessage(messages, context),
+    MaxTokens:   maxTokens,
 		Temperature: temperature,
 	}
 
diff --git a/RequestOpenai.go b/RequestOpenai.go
index dc7b6bd..ad1f950 100644
--- a/RequestOpenai.go
+++ b/RequestOpenai.go
@@ -14,6 +14,7 @@ import (
 type OpenaiChatCompletionRequest struct {
 	Model       string           `json:"model"`
 	Messages    []RequestMessage `json:"messages"`
+  MaxTokens   int              `json:"max_tokens"`
 	Temperature float64          `json:"temperature"`
 }
 
@@ -41,7 +42,7 @@ type OpenaiChoice struct {
 func addOpenaiMessage(c *fiber.Ctx, llm LLM, selected bool) edgedb.UUID {
 	Messages := getAllSelectedMessages(c)
 
-	chatCompletion, err := RequestOpenai(c, llm.Model.ModelID, Messages, float64(llm.Temperature), llm.Context)
+	chatCompletion, err := RequestOpenai(c, llm.Model.ModelID, Messages, float64(llm.Temperature), llm.Context, int(llm.MaxToken))
 	if err != nil {
     fmt.Println("Error requesting OpenAI: ", err)
     id := insertBotMessage(c, "Error requesting OpenAI, model may not be available anymore. Better error message in development.", selected, llm.ID)
@@ -110,7 +111,7 @@ func TestOpenaiKey(apiKey string) bool {
 	return true
 }
 
-func RequestOpenai(c *fiber.Ctx, model string, messages []Message, temperature float64, context string) (OpenaiChatCompletionResponse, error) {
+func RequestOpenai(c *fiber.Ctx, model string, messages []Message, temperature float64, context string, maxTokens int) (OpenaiChatCompletionResponse, error) {
 	var apiKey string
 	err := edgeGlobalClient.WithGlobals(map[string]interface{}{"ext::auth::client_token": c.Cookies("jade-edgedb-auth-token")}).QuerySingle(edgeCtx, `
 		with
@@ -130,6 +131,7 @@ func RequestOpenai(c *fiber.Ctx, model string, messages []Message, temperature f
 	requestBody := OpenaiChatCompletionRequest{
 		Model:       model,
 		Messages:    Message2RequestMessage(messages, context),
+    MaxTokens:   maxTokens,
 		Temperature: temperature,
 	}
 
diff --git a/RequestPerplexity.go b/RequestPerplexity.go
index 22d1d77..98e5b20 100644
--- a/RequestPerplexity.go
+++ b/RequestPerplexity.go
@@ -15,6 +15,7 @@ import (
 type PerplexityChatCompletionRequest struct {
 	Model       string           `json:"model"`
 	Messages    []RequestMessage `json:"messages"`
+  MaxTokens   int              `json:"max_tokens"`
 	Temperature float64          `json:"temperature"`
 }
 
@@ -42,7 +43,7 @@ type PerplexityChoice struct {
 func addPerplexityMessage(c *fiber.Ctx, llm LLM, selected bool) edgedb.UUID {
 	Messages := getAllSelectedMessages(c)
 
-	chatCompletion, err := RequestPerplexity(c, llm.Model.ModelID, Messages, float64(llm.Temperature), llm.Context)
+	chatCompletion, err := RequestPerplexity(c, llm.Model.ModelID, Messages, float64(llm.Temperature), llm.Context, int(llm.MaxToken))
 	if err != nil {
     fmt.Println("Error requesting Perplexity: ", err)
     id := insertBotMessage(c, "Error requesting Perplexity, model may not be available anymore. Better error message in development.", selected, llm.ID)
@@ -111,7 +112,7 @@ func TestPerplexityKey(apiKey string) bool {
 	return true
 }
 
-func RequestPerplexity(c *fiber.Ctx, model string, messages []Message, temperature float64, context string) (PerplexityChatCompletionResponse, error) {
+func RequestPerplexity(c *fiber.Ctx, model string, messages []Message, temperature float64, context string, maxTokens int) (PerplexityChatCompletionResponse, error) {
 	var apiKey string
 	err := edgeGlobalClient.WithGlobals(map[string]interface{}{"ext::auth::client_token": c.Cookies("jade-edgedb-auth-token")}).QuerySingle(edgeCtx, `
 		with
@@ -131,6 +132,7 @@ func RequestPerplexity(c *fiber.Ctx, model string, messages []Message, temperatu
 	requestBody := PerplexityChatCompletionRequest{
 		Model:       model,
 		Messages:    Message2RequestMessage(messages, context),
+    MaxTokens:   maxTokens,
 		Temperature: temperature,
 	}