How to resolve LangExtract OpenAI setup errors when switching from Gemini

Feb 12, 2026

Cowrie

Dev @ Bswen

Problem

When I tried using LangExtract with OpenAI’s GPT models after setting it up with Gemini, I got these errors:

Traceback (most recent call last):
  File "/path/to/test.py", line 5, in <module>
    import langextract as lx
  File "/path/to/langextract/__init__.py", line 10, in <module>
    from .extract import extract
  File "/path/to/langextract/extract.py", line 15, in <module>
    from openai import OpenAI
ModuleNotFoundError: No module named 'openai'

Then after installing the missing module, I got authentication errors:

openai.AuthenticationError: Error code: 401 - Incorrect API key provided

And after fixing the API key, the output parsing failed:

json.decoder.JSONDecodeError: Expecting value: line 1 column 1 (char 0)

Environment

• Python 3.11
• LangExtract 0.1.0
• OpenAI GPT-4o
• macOS 14.5

What happened?

I was using LangExtract with Gemini and it worked fine. Here’s my working Gemini setup:

gemini_example.py

import langextract as lx
import os

# Set API key
os.environ['LANGEXTRACT_API_KEY'] = 'my-gemini-key'

result = lx.extract(
    text="Patient prescribed Amoxicillin 500mg twice daily.",
    prompt_description="Extract medications, dosages, and frequency",
    examples=[
        lx.data.ExampleData(
            text="Take Aspirin 100mg orally once daily",
            extractions=[
                lx.data.Extraction(
                    extraction_class="medication",
                    extraction_text="Aspirin",
                    attributes={
                        "dosage": "100mg",
                        "route": "orally",
                        "frequency": "once daily"
                    }
                )
            ]
        )
    ],
    model_id="gemini-1.5-flash"
)

I can explain the key parts:

• Using LANGEXTRACT_API_KEY environment variable
• Default parameters work for Gemini
• No special installation needed

So I tried switching to OpenAI by changing the model_id:

os.environ['LANGEXTRACT_API_KEY'] = 'sk-proj-...'  # OpenAI key

result = lx.extract(
    text="Patient prescribed Amoxicillin 500mg twice daily.",
    prompt_description="Extract medications, dosages, and frequency",
    examples=[...],  # Same examples
    model_id="gpt-4o"  # Changed to OpenAI
)

But when I ran this, I got the ModuleNotFoundError: No module named 'openai'.

How to solve it?

Solution #1: Install OpenAI Extra

I checked the LangExtract documentation and found OpenAI support is an optional dependency. I tried installing it:

Terminal

pip install langextract[openai]

Or for development installs:

Terminal

pip install -e ".[openai]"

This fixed the first error.

Solution #2: Fix API Key Variable

After installing the OpenAI dependency, I ran the code again and got AuthenticationError: 401.

I checked the LangExtract source code and found OpenAI uses a different environment variable. I changed from LANGEXTRACT_API_KEY to OPENAI_API_KEY:

openai_example_fixed.py

import langextract as lx
import os

# WRONG for OpenAI
# os.environ['LANGEXTRACT_API_KEY'] = 'sk-proj-...'

# CORRECT for OpenAI
os.environ['OPENAI_API_KEY'] = 'sk-proj-...'

Solution #3: Disable Schema Constraints

After fixing the API key, the code ran but returned empty extractions or JSON parsing errors. I tried checking the actual LLM output and found OpenAI was returning unfenced JSON.

LangExtract has two important parameters for OpenAI:

result = lx.extract(
    text="Patient prescribed Amoxicillin 500mg twice daily.",
    prompt_description="Extract medications, dosages, and frequency",
    examples=[...],
    model_id="gpt-4o",
    api_key=os.environ['OPENAI_API_KEY'],
    use_schema_constraints=False,  # OpenAI doesn't support this
    fence_output=True  # Wrap JSON in markdown code blocks
)

The key changes:

• use_schema_constraints=False: OpenAI doesn’t support structured output constraints the way Gemini does
• fence_output=True: Tells the LLM to wrap JSON in markdown code blocks for reliable parsing

Now test again:

print(f"Extracted: {result.extractions}")
# Output: Extracted: [Extraction(extraction_class='medication', extraction_text='Amoxicillin', attributes={'dosage': '500mg', 'frequency': 'twice daily'})]

You can see that I succeeded to extract the medication data with OpenAI.

The reason

I think the key reason for these errors is that LangExtract defaults are optimized for Gemini, not OpenAI:

1. Missing dependency: OpenAI support is opt-in via [openai] extra, not installed by default
2. API key mismatch: OpenAI SDK looks for OPENAI_API_KEY, not LANGEXTRACT_API_KEY
3. Schema constraints: Gemini supports structured output constraints via schema validation, but OpenAI doesn’t have this feature
4. Output fencing: GPT models often output raw JSON without markdown code blocks, which breaks LangExtract’s parser

Here’s the API key priority matrix:

Provider	Primary Env Var	Fallback Env Var	Notes
Gemini	GEMINI_API_KEY	LANGEXTRACT_API_KEY	Default
OpenAI	OPENAI_API_KEY	LANGEXTRACT_API_KEY	Requires [openai] install
Ollama	(none needed)	(none needed)	Local only

The fence_output parameter is critical for OpenAI. Without it:

GPT output: {"entity": "value"}  # Raw JSON

With fence_output=True:

GPT output: ```json
{"entity": "value"}

LangExtract's parser knows to extract JSON from the code block.

## Complete Working Example

Here's a complete example with error handling and proper configuration:

```python title="openai_complete_example.py"
import langextract as lx
import os
from dotenv import load_dotenv

# Load environment variables
load_dotenv()

# Verify API key is set
api_key = os.environ.get('OPENAI_API_KEY')
if not api_key:
    raise ValueError(
        "OPENAI_API_KEY not set. "
        "OpenAI requires specific env variable, not LANGEXTRACT_API_KEY."
    )

# Configure OpenAI-specific parameters
result = lx.extract(
    text="Patient prescribed Amoxicillin 500mg twice daily.",
    prompt_description="Extract medications, dosages, and frequency",
    examples=[
        lx.data.ExampleData(
            text="Take Aspirin 100mg orally once daily",
            extractions=[
                lx.data.Extraction(
                    extraction_class="medication",
                    extraction_text="Aspirin",
                    attributes={
                        "dosage": "100mg",
                        "route": "orally",
                        "frequency": "once daily"
                    }
                )
            ]
        )
    ],
    model_id="gpt-4o",
    api_key=api_key,
    fence_output=True,  # Required for OpenAI
    use_schema_constraints=False  # Not supported for OpenAI
)

print(f"Extracted: {result.extractions}")

Summary

In this post, I showed how to resolve LangExtract OpenAI setup errors when switching from Gemini. The key point is OpenAI requires the [openai] extra installation, OPENAI_API_KEY environment variable, and specific parameters (fence_output=True, use_schema_constraints=False) to work properly with LangExtract.

Final Words + More Resources

My intention with this article was to help others share my knowledge and experience. If you want to contact me, you can contact by email: Email me

Here are also the most important links from this article along with some further resources that will help you in this scope:

• 👨‍💻 LangExtract Official Documentation
• 👨‍💻 OpenAI API Documentation
• 👨‍💻 LangExtract GitHub Issues - OpenAI Provider

Oh, and if you found these resources useful, don’t forget to support me by starring the repo on GitHub!

Comments

Tags:

• llm
• ai
• python

How to resolve LangExtract Ollama timeout and connection errors

Marimo vs Jupyter: Why Switch for Python Data Science?