Code
T7107 answer-cut-off
The answer of the model was cut off: its server said that it ended because it reached its length. The call fails with the status 500, and is given to no second model.
Details
Of a stream with a shape, the parts that were given before it stay given, and the stream ends with this code.
The fix
Ask for a shorter answer, or give the model more room through its options, in the words of its provider.
TypeScript
export const writer = model('writer', { provider: 'anthropic', name: 'claude-sonnet-5-5', key: secret('ANTHROPIC_KEY'), options: { max_tokens: 16000 } })