Fix Azure OpenAI deployment not found error
deployment not found error occurs when the model parameter in AzureOpenAI calls does not match the deployment name configured in your Azure OpenAI resource. Use the AzureOpenAI client with the exact deployment name (not the model name) and ensure azure_endpoint and api_version are set correctly.config_error model parameter when calling AzureOpenAI.chat.completions.create().Why this happens
This error is triggered when your code uses the Azure OpenAI SDK but specifies a model parameter that is not the deployment name you created in the Azure portal. Azure OpenAI requires you to deploy a model under a custom deployment name and then use that deployment name in API calls, not the base model name like gpt-4o.
Typical error message:
azure.core.exceptions.ResourceNotFoundError: Deployment not found
Example of incorrect code causing this error:
from openai import AzureOpenAI
import os
client = AzureOpenAI(
api_key=os.environ["AZURE_OPENAI_API_KEY"],
azure_endpoint=os.environ["AZURE_OPENAI_ENDPOINT"],
api_version="2024-02-01"
)
response = client.chat.completions.create(
model="gpt-4o", # Incorrect: should be deployment name
messages=[{"role": "user", "content": "Hello"}]
)
print(response.choices[0].message.content)azure.core.exceptions.ResourceNotFoundError: Deployment not found
The fix
Use the exact deployment name you configured in the Azure portal for your OpenAI resource as the model parameter. This deployment name is case-sensitive and often looks like gpt4o-deployment or a custom string you chose.
Ensure you instantiate AzureOpenAI with the correct azure_endpoint and api_version. Then call chat.completions.create() with the deployment name.
from openai import AzureOpenAI
import os
client = AzureOpenAI(
api_key=os.environ["AZURE_OPENAI_API_KEY"],
azure_endpoint=os.environ["AZURE_OPENAI_ENDPOINT"],
api_version="2024-02-01"
)
response = client.chat.completions.create(
model=os.environ["AZURE_OPENAI_DEPLOYMENT"], # Correct deployment name
messages=[{"role": "user", "content": "Hello"}]
)
print(response.choices[0].message.content)Hello
Preventing it in production
- Validate your deployment names by listing them in the Azure portal or via Azure CLI before using them in code.
- Use environment variables for deployment names to avoid hardcoding and typos.
- Implement retry logic with exponential backoff for transient errors.
- Log API errors clearly to detect misconfiguration early.
- Keep
azure_endpointandapi_versionupdated per Azure documentation.
Key Takeaways
- Always use the Azure deployment name, not the base model name, in
modelparameter. - Configure
AzureOpenAIclient with correctazure_endpointandapi_version. - Use environment variables for deployment names to avoid typos and ease updates.