FireworksAIJSONParseError
fireworks_ai.errors.FireworksAIJSONParseError
Stack trace
fireworks_ai.errors.FireworksAIJSONParseError: Failed to parse JSON output from LLM: Expecting value: line 1 column 1 (char 0)
File "/app/fireworks_ai/parser.py", line 45, in parse_json_mode
parsed = json.loads(raw_response) # <-- error here
File "/usr/lib/python3.9/json/__init__.py", line 346, in loads
return _default_decoder.decode(s)
File "/usr/lib/python3.9/json/decoder.py", line 337, in decode
obj, end = self.raw_decode(s, idx=_w(s, 0).end())
File "/usr/lib/python3.9/json/decoder.py", line 355, in raw_decode
raise JSONDecodeError("Expecting value", s, err.value) from None Why it happens
Fireworks AI expects the LLM to return a strictly valid JSON string when operating in JSON mode. If the LLM returns extra text, markdown fences, or malformed JSON, the parser raises this error because json.loads() cannot decode the response.
Detection
Log or capture the raw LLM response before parsing and validate it against JSON syntax; catching FireworksAIJSONParseError allows you to inspect malformed outputs before crashing.
Causes & fixes
LLM response includes markdown fences or extra explanatory text around JSON
Modify the prompt to instruct the LLM to return only raw JSON without markdown or extra text, or preprocess the response to strip fences before parsing.
LLM output contains trailing commas or invalid JSON syntax
Use a JSON sanitizer or stricter prompt instructions to ensure the LLM outputs valid JSON without trailing commas or syntax errors.
Using a base model that does not reliably follow JSON output instructions
Switch to a model fine-tuned for structured JSON output or use Fireworks AI’s recommended models that enforce JSON compliance.
Code: broken vs fixed
from fireworks_ai import FireworksAI
client = FireworksAI(api_key="sk-incorrect")
response = client.generate(prompt="Return JSON mode output")
parsed = client.parse_json_mode(response) # This line raises FireworksAIJSONParseError
print(parsed)import os
from fireworks_ai import FireworksAI
client = FireworksAI(api_key=os.environ["FIREWORKS_API_KEY"])
prompt = "Return ONLY raw JSON without markdown fences or extra text."
response = client.generate(prompt=prompt)
# Strip markdown fences if present
clean_response = response.strip().strip('`')
parsed = client.parse_json_mode(clean_response) # Fixed: cleaned input for JSON parsing
print(parsed)Workaround
Catch FireworksAIJSONParseError, extract JSON substring using regex from the raw response, then parse with json.loads() as a fallback to handle malformed outputs.
Prevention
Use strict prompt engineering to enforce JSON-only output and validate responses with schema validators or Fireworks AI’s built-in JSON output modes to avoid parsing errors.