Fix Python TypeError: Object of type 'int64' is not JSON serializable

The Python TypeError: Object of type 'int64' is not JSON serializable error message occurs when you try to convert an int64 object to a JSON string.

To fix this error, you need to convert the int64 type to a JSON-serializable type(such as int).

This error commonly happens when you use the NumPy module in your Python code.

Suppose you call the json.dumps() function on a NumPy int64 object as shown below:

import numpy as np
import json

score = np.int64(90)

json_str = json.dumps(score)  # ❌

The JSON dump() and dumps() methods can only convert Python native data types, such as:

• dict
• list
• str
• int
• float
• bool
• None

Since int64 is a data type created by NumPy, the json module doesn’t know how to handle it.

This is why Python returns the following error:

Traceback (most recent call last):
  File ...
TypeError: Object of type int64 is not JSON serializable

To solve this error, you can convert the int64 object into a native Python int object using the int() construct.

See the example below:

import numpy as np
import json

score = np.int64(90)
int_value = int(score)

json_str = json.dumps(int_value)  # ✅
print(json_str)  # 90

If you have lots of Numpy int64 variables in your code, then converting each of the variables into int before calling json.dumps() can be tedious.

Instead of manually converting the int64 values, you can extend the JSONEncoder class and instruct the class on how to convert NumPy custom objects into serializable objects.

Here’s how you create a custom JSON encoder in Python:

import json
import numpy as np

# Extend the JSONEncoder class
class NpEncoder(json.JSONEncoder):
    def default(self, obj):
        if isinstance(obj, np.integer):
            return int(obj)
        if isinstance(obj, np.floating):
            return float(obj)
        if isinstance(obj, np.ndarray):
            return obj.tolist()
        return json.JSONEncoder.default(self, obj)

The NpEncoder class above extends the JSONEncoder class from the json module.

The default() method is called whenever the encoder encounters an object that it does not know how to serialize. Otherwise, the encoder will run without calling this method.

Because you want to handle NumPy objects, you need to check whether the object passed is an instance of np.integer, np.floating, or np.ndarray object.

The object is then converted into an int, float, or list so that it can be processed by the encoder.

To use this extended encoder, you need to pass it as the cls argument into json.dumps() method.

See the highlighted line below:

score = np.int64(90)

json_str = json.dumps(score, cls=NpEncoder)  # ✅
print(json_str)

As you can see, using the NpEncoder class enables the json.dumps() method to convert an int64 object.

The extended if check also allows you to convert a NumPy ndarray type, avoiding the ndarray is not JSON serializable error.

Conclusion

To summarize, you encounter the Python TypeError: Object of type 'int64' is not JSON serializable error message when you try to convert an int64 object into a JSON string.

To fix this error, you can convert the int64 data type to an int, or you can create a custom class that extends the JSON.encoder class.

If you’re only dealing with a small amount of NumPy data, then converting the type directly is enough.

Otherwise, extending the JSON encoder class is a great solution as you can avoid repeating yourself.