diff --git a/metagpt/utils/token_counter.py b/metagpt/utils/token_counter.py
index a57871de7..d68b9e752 100644
--- a/metagpt/utils/token_counter.py
+++ b/metagpt/utils/token_counter.py
@@ -406,13 +406,17 @@ SPARK_TOKENS = {
 def count_message_tokens(messages, model="gpt-3.5-turbo-0125"):
     """Return the number of tokens used by a list of messages."""
     if "claude" in model:
-        # rough estimation for models newer than claude-2.1
-        vo = anthropic.Client()
-        num_tokens = 0
-        for message in messages:
-            for key, value in message.items():
-                num_tokens += vo.count_tokens(str(value))
-        return num_tokens
+        # rough estimation for models newer than claude-2.1, needs api_key or auth_token
+        ac = anthropic.Client()
+        system_prompt = ""
+        new_messages = []
+        for msg in messages:
+            if msg.get("role") == "system":
+                system_prompt = msg.get("content")
+            else:
+                new_messages.append(msg)
+        num_tokens = ac.beta.messages.count_tokens(messages=new_messages, model=model, system=system_prompt)
+        return num_tokens.get("input_tokens", 0)
     try:
         encoding = tiktoken.encoding_for_model(model)
     except KeyError:
diff --git a/setup.py b/setup.py
index 658c82219..1fe74b64d 100644
--- a/setup.py
+++ b/setup.py
@@ -98,7 +98,7 @@ extras_require["android_assistant"] = [
 
 setup(
     name="metagpt",
-    version="0.8.1",
+    version="1.0.0",
     description="The Multi-Agent Framework",
     long_description=long_description,
     long_description_content_type="text/markdown",