ValidationError
pydantic.error_wrappers.ValidationError
Stack trace
pydantic.error_wrappers.ValidationError: 1 validation error for OutputModel field_name field required (type=value_error.missing) During handling of the above exception, the Anthropic client raised a ValidationError when parsing the structured output.
Why it happens
Anthropic's structured output expects the LLM to return JSON strictly matching the Pydantic model schema. If the LLM adds extra text, markdown fences, or omits required fields, Pydantic raises a ValidationError. This often happens if the prompt does not enforce strict JSON-only output or if the model is not instruction-tuned.
Detection
Catch pydantic.ValidationError exceptions during output parsing and log the raw LLM response to identify mismatches before the app crashes.
Causes & fixes
LLM response includes markdown fences or extra text around the JSON output
Modify the prompt to instruct the model to return only raw JSON without markdown fences, or use a parser that strips fences automatically.
Pydantic model fields do not exactly match the keys in the LLM's JSON output
Ensure the Pydantic model field names and types exactly match the expected JSON keys and data types returned by the LLM.
Using a base Anthropic model that does not reliably follow structured output instructions
Switch to an instruction-tuned Anthropic model like claude-3-5-haiku-20241022 that better respects output format constraints.
Code: broken vs fixed
import os
from anthropic import Anthropic
from pydantic import BaseModel
class OutputModel(BaseModel):
field_name: str
client = Anthropic(api_key=os.environ['ANTHROPIC_API_KEY'])
response = client.messages.create(
model='claude-3-5-haiku-20241022',
system='You are a helpful assistant.',
messages=[{'role': 'user', 'content': 'Return data as JSON with field_name.'}]
)
# This line raises ValidationError because response includes markdown fences
output = OutputModel.parse_raw(response['completion']) # Error hereimport os
from anthropic import Anthropic
from pydantic import BaseModel
class OutputModel(BaseModel):
field_name: str
client = Anthropic(api_key=os.environ['ANTHROPIC_API_KEY'])
prompt = (
'Return ONLY raw JSON with field_name, no markdown fences or extra text.'
)
response = client.messages.create(
model='claude-3-5-haiku-20241022',
system='You are a helpful assistant.',
messages=[{'role': 'user', 'content': prompt}]
)
output = OutputModel.parse_raw(response['completion']) # Fixed: prompt enforces raw JSON
print(output)Workaround
Wrap the Pydantic parse call in try/except ValidationError, then extract JSON from the raw string using regex and parse it manually with json.loads() as a fallback.
Prevention
Use Anthropic's structured output features or tool use to enforce schema-valid JSON responses at the API level, avoiding fragile post-processing parsing.