In Keras, the API allows for more complex models with non-linear topology, shared layers, and even multiple inputs or outputs.

The API in Keras that allows for more complex models with non-linear topology, shared layers, and even multiple inputs or outputs is the Functional API.

Here are the steps to use it:

1. Import the necessary modules:

from keras.models import Model
from keras.layers import Input, Dense

2. Define the input layer. This returns a tensor:

inputs = Input(shape=(10,))

3. Create a new node in the graph of layers by calling a layer on this inputs object:

x = Dense(64, activation='relu')(inputs)

4. Continue building the model by adding more layers:

x = Dense(64, activation='relu')(x)
predictions = Dense(10, activation='softmax')(x)

5. Create the model:

model = Model(inputs=inputs, outputs=predictions)

6. Now, compile the model:

model.compile(optimizer='rmsprop',
              loss='categorical_crossentropy',
              metrics=['accuracy'])

7. Finally, train the model with some data:

model.fit(data, labels)

This is a basic example of how to use the Functional API in Keras. It allows for more flexibility in designing complex models compared to the Sequential API.