Forcing JSON the old way

An extraction pipeline written against an older model has started returning 400s after a model upgrade. It worked unchanged for a year.

response = client.messages.create(
    model="claude-opus-5",
    max_tokens=4096,
    messages=[
        {"role": "user", "content": f"Extract the contact details:\n{text}"},
        {"role": "assistant", "content": '{"name": "'},
    ],
)