diff --git a/pkg/llm/client/glm-6b/client.go b/pkg/llm/client/glm-6b/client.go
index 0d915ed..52d05d9 100644
--- a/pkg/llm/client/glm-6b/client.go
+++ b/pkg/llm/client/glm-6b/client.go
@@ -27,15 +27,18 @@ import (
 
 	"github.com/basenana/friday/pkg/llm"
 	"github.com/basenana/friday/pkg/llm/prompts"
+	"github.com/basenana/friday/pkg/utils/logger"
 )
 
 type GLM struct {
+	log     logger.Logger
 	baseUri string
 }
 
 func NewGLM(uri string) llm.LLM {
 	return &GLM{
 		baseUri: uri,
+		log:     logger.NewLogger("glm"),
 	}
 }
 
@@ -62,6 +65,7 @@ func (o *GLM) request(path string, method string, body io.Reader) ([]byte, error
 	if resp.StatusCode != 200 {
 		return nil, fmt.Errorf("fail to call glm-6b, status code error: %d", resp.StatusCode)
 	}
+	o.log.Debugf("openai response: %s", respBody)
 	return respBody, nil
 }
 
diff --git a/pkg/llm/client/openai/v1/chat.go b/pkg/llm/client/openai/v1/chat.go
index 3f165fe..2466d32 100644
--- a/pkg/llm/client/openai/v1/chat.go
+++ b/pkg/llm/client/openai/v1/chat.go
@@ -19,8 +19,6 @@ package v1
 import (
 	"context"
 	"encoding/json"
-	"strings"
-	"time"
 
 	"github.com/basenana/friday/pkg/llm/prompts"
 )
@@ -41,17 +39,7 @@ type ChatChoice struct {
 }
 
 func (o *OpenAIV1) Chat(ctx context.Context, prompt prompts.PromptTemplate, parameters map[string]string) ([]string, error) {
-	answer, err := o.chat(ctx, prompt, parameters)
-	if err != nil {
-		errMsg := err.Error()
-		if strings.Contains(errMsg, "rate_limit_exceeded") || strings.Contains(errMsg, "Rate limit reached") {
-			o.log.Warn("meets rate limit exceeded, sleep 30 seconds and retry")
-			time.Sleep(time.Duration(30) * time.Second)
-			return o.chat(ctx, prompt, parameters)
-		}
-		return nil, err
-	}
-	return answer, err
+	return o.chat(ctx, prompt, parameters)
 }
 
 func (o *OpenAIV1) chat(ctx context.Context, prompt prompts.PromptTemplate, parameters map[string]string) ([]string, error) {
@@ -62,7 +50,6 @@ func (o *OpenAIV1) chat(ctx context.Context, prompt prompts.PromptTemplate, para
 	if err != nil {
 		return nil, err
 	}
-	o.log.Debugf("final prompt: %s", p)
 
 	data := map[string]interface{}{
 		"model":             model,
diff --git a/pkg/llm/client/openai/v1/client.go b/pkg/llm/client/openai/v1/client.go
index 90a6cda..77b41cf 100644
--- a/pkg/llm/client/openai/v1/client.go
+++ b/pkg/llm/client/openai/v1/client.go
@@ -90,7 +90,6 @@ func (o *OpenAIV1) request(ctx context.Context, path string, method string, data
 			req.Header.Set("Content-Type", "application/json; charset=utf-8")
 			req.Header.Set("Authorization", fmt.Sprintf("Bearer %s", o.key))
 
-			o.log.Debugf("request: %s", uri)
 			client := &http.Client{}
 			resp, err := client.Do(req)
 			if err != nil {
@@ -108,7 +107,7 @@ func (o *OpenAIV1) request(ctx context.Context, path string, method string, data
 				time.Sleep(time.Second * 30)
 				continue
 			}
-			//o.log.Debugf("response: %s", respBody)
+			o.log.Debugf("openai response: %s", respBody)
 			return respBody, nil
 		}
 	}
diff --git a/pkg/llm/client/openai/v1/completion.go b/pkg/llm/client/openai/v1/completion.go
index caadb97..3e4b3e5 100644
--- a/pkg/llm/client/openai/v1/completion.go
+++ b/pkg/llm/client/openai/v1/completion.go
@@ -19,8 +19,6 @@ package v1
 import (
 	"context"
 	"encoding/json"
-	"strings"
-	"time"
 
 	"github.com/basenana/friday/pkg/llm/prompts"
 )
@@ -42,17 +40,7 @@ type Choice struct {
 }
 
 func (o *OpenAIV1) Completion(ctx context.Context, prompt prompts.PromptTemplate, parameters map[string]string) ([]string, error) {
-	answer, err := o.completion(ctx, prompt, parameters)
-	if err != nil {
-		errMsg := err.Error()
-		if strings.Contains(errMsg, "rate_limit_exceeded") {
-			o.log.Warn("meets rate limit exceeded, sleep 30 seconds and retry")
-			time.Sleep(time.Duration(30) * time.Second)
-			return o.completion(ctx, prompt, parameters)
-		}
-		return nil, err
-	}
-	return answer, err
+	return o.completion(ctx, prompt, parameters)
 }
 
 func (o *OpenAIV1) completion(ctx context.Context, prompt prompts.PromptTemplate, parameters map[string]string) ([]string, error) {
diff --git a/pkg/llm/client/openai/v1/embedding.go b/pkg/llm/client/openai/v1/embedding.go
index 2437f81..b90356c 100644
--- a/pkg/llm/client/openai/v1/embedding.go
+++ b/pkg/llm/client/openai/v1/embedding.go
@@ -19,8 +19,6 @@ package v1
 import (
 	"context"
 	"encoding/json"
-	"strings"
-	"time"
 )
 
 type EmbeddingResult struct {
@@ -46,17 +44,7 @@ type Usage struct {
 }
 
 func (o *OpenAIV1) Embedding(ctx context.Context, doc string) (*EmbeddingResult, error) {
-	answer, err := o.embedding(ctx, doc)
-	if err != nil {
-		errMsg := err.Error()
-		if strings.Contains(errMsg, "rate_limit_exceeded") {
-			o.log.Warn("meets rate limit exceeded, sleep 30 seconds and retry")
-			time.Sleep(time.Duration(30) * time.Second)
-			return o.embedding(ctx, doc)
-		}
-		return nil, err
-	}
-	return answer, err
+	return o.embedding(ctx, doc)
 }
 
 func (o *OpenAIV1) embedding(ctx context.Context, doc string) (*EmbeddingResult, error) {