Update model_prices_and_context_window.json

[MohammedSamirMahmoud]

added custom naming for different models supported ending with
-gs for Azure global standard deployments
-dzs for Azure Data Zone Standard deployments

Title

Relevant issues

Pre-Submission checklist

Please complete all items before asking a LiteLLM maintainer to review your PR

• I have Added testing in the tests/litellm/ directory, Adding at least 1 test is a hard requirement - see details
• I have added a screenshot of my new test passing locally
• My PR passes all unit tests on make test-unit
• My PR's scope is as isolated as possible, it only solves 1 specific problem

Type

🆕 New Feature
🐛 Bug Fix
🧹 Refactoring
📖 Documentation
🚄 Infrastructure
✅ Test

Changes

[CLAassistant]

All committers have signed the CLA.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Preview	Comments	Updated (UTC)
litellm	Ready	Preview	Comment	Aug 20, 2025 10:20am

[krrishdholakia]

@MohammedSamirMahmoud the name change would break cost tracking based on returned azure model

i recommend having global standard, be the 'default' pricing (no special -gs)

[krrishdholakia]

is the pricing the same across all data zones?

[krrishdholakia]

if not - what does -dzs mean? and how would it account for regional differences

[MohammedSamirMahmoud]

Hey @krrishdholakia ,

So during the usage of litellm as a proxy server.
For Azure models we are setting up the following naming convention as a standard

azure/gpt-4.1 models:

gpt-41-gs --> Stands for Azure GPT 4.1 models deployed as Global standard
gpt-41-dzs --> Stands for Azure GPT 4.1 models deployed as Data Zone Standard (DZS)
gpt-41--> Stands for Azure GPT 4.1 models deployed as Global standard

gpt-5-dzs --> Stands for Azure GPT 5 models deployed as Data Zone Standard (DZS)
gpt-5-mini-dzs--> Stands for Azure GPT 5 models deployed as Data Zone Standard (DZS)

Now while calling these models as completions--> we get alot of informative headers from LiteLLM including model_region, model_tokens, model_cost, etc.

However when doing this using the responses endpoint, no headers are retrieved at all!

Typical error when starting LiteLLM Proxy Server:

{"message": "checking potential_model_names in litellm.model_cost: {'split_model': 'gpt-5-dzs', 'combined_model_name': 'azure/gpt-5-dzs', 'stripped_model_name': 'azure/gpt-5-dzs', 'combined_stripped_model_name': 'azure/gpt-5-dzs', 'custom_llm_provider': 'azure'}", "level": "DEBUG", "timestamp": "2025-08-21T13:19:18.045934"}

{"message": "Error getting model info: This model isn't mapped yet. Add it here - https://github.com/BerriAI/litellm/blob/main/model_prices_and_context_window.json", "level": "DEBUG", "timestamp": "2025-08-21T13:19:18.045948"}
{"message": "added/updated model=azure/gpt-5-dzs in litellm.model_cost: azure/gpt-5-dzs", "level": "DEBUG", "timestamp": "2025-08-21T13:19:18.045969"}
{"message": "checking potential_model_names in litellm.model_cost: {'split_model': '69260937c7da004ba35663898e26638c0074cdfac2bc2a437189a78a2562c3bc', 'combined_model_name': '69260937c7da004ba35663898e26638c0074cdfac2bc2a437189a78a2562c3bc', 'stripped_model_name': '69260937c7da004ba35663898e26638c0074cdfac2bc2a437189a78a2562c3bc', 'combined_stripped_model_name': '69260937c7da004ba35663898e26638c0074cdfac2bc2a437189a78a2562c3bc', 'custom_llm_provider': None}", "level": "DEBUG", "timestamp": "2025-08-21T13:19:18.046139"}
{"message": "Error getting model info: This model isn't mapped yet. Add it here - https://github.com/BerriAI/litellm/blob/main/model_prices_and_context_window.json", "level": "DEBUG", "timestamp": "2025-08-21T13:19:18.046157"}
{"message": "added/updated model=69260937c7da004ba35663898e26638c0074cdfac2bc2a437189a78a2562c3bc in litellm.model_cost: 69260937c7da004ba35663898e26638c0074cdfac2bc2a437189a78a2562c3bc", "level": "DEBUG", "timestamp": "2025-08-21T13:19:18.046170"}
{"message": "checking potential_model_names in litellm.model_cost: {'split_model': 'responses/gpt-5-dzs', 'combined_model_name': 'azure/responses/gpt-5-dzs', 'stripped_model_name': 'azure/responses/gpt-5-dzs', 'combined_stripped_model_name': 'azure/responses/gpt-5-dzs', 'custom_llm_provider': 'azure'}", "level": "DEBUG", "timestamp": "2025-08-21T13:19:18.046202"}
{"message": "Error getting model info: This model isn't mapped yet. Add it here - https://github.com/BerriAI/litellm/blob/main/model_prices_and_context_window.json", "level": "DEBUG", "timestamp": "2025-08-21T13:19:18.046215"}
{"message": "added/updated model=azure/responses/gpt-5-dzs in litellm.model_cost: azure/responses/gpt-5-dzs", "level": "DEBUG", "timestamp": "2025-08-21T13:19:18.046227"}
{"message": "checking potential_model_names in litellm.model_cost: {'split_model': 'ce064fe813df870a53d7eaccbc5bac1fd09ea19673047cbfae522d57a33c7278', 'combined_model_name': 'ce064fe813df870a53d7eaccbc5bac1fd09ea19673047cbfae522d57a33c7278', 'stripped_model_name': 'ce064fe813df870a53d7eaccbc5bac1fd09ea19673047cbfae522d57a33c7278', 'combined_stripped_model_name': 'ce064fe813df870a53d7eaccbc5bac1fd09ea19673047cbfae522d57a33c7278', 'custom_llm_provider': None}", "level": "DEBUG", "timestamp": "2025-08-21T13:19:18.046398"}
{"message": "Error getting model info: This model isn't mapped yet. Add it here - https://github.com/BerriAI/litellm/blob/main/model_prices_and_context_window.json", "level": "DEBUG", "timestamp": "2025-08-21T13:19:18.046415"}
{"message": "added/updated model=ce064fe813df870a53d7eaccbc5bac1fd09ea19673047cbfae522d57a33c7278 in litellm.model_cost: ce064fe813df870a53d7eaccbc5bac1fd09ea19673047cbfae522d57a33c7278", "level": "DEBUG", "timestamp": "2025-08-21T13:19:18.046429"}
{"message": "checking potential_model_names in litellm.model_cost: {'split_model': 'gpt-5-mini-dzs', 'combined_model_name': 'azure/gpt-5-mini-dzs', 'stripped_model_name': 'azure/gpt-5-mini-dzs', 'combined_stripped_model_name': 'azure/gpt-5-mini-dzs', 'custom_llm_provider': 'azure'}", "level": "DEBUG", "timestamp": "2025-08-21T13:19:18.046459"}
{"message": "Error getting model info: This model isn't mapped yet. Add it here - https://github.com/BerriAI/litellm/blob/main/model_prices_and_context_window.json", "level": "DEBUG", "timestamp": "2025-08-21T13:19:18.046473"}
{"message": "added/updated model=azure/gpt-5-mini-dzs in litellm.model_cost: azure/gpt-5-mini-dzs", "level": "DEBUG", "timestamp": "2025-08-21T13:19:18.046484"}

This typical behavior occurs only with responses endpoint, whenever the deployment-name is not exactly as your model maps, when i tried to create a new deployment following just whatever inside your model_maps --> azure/gpt-5, it worked fine! and retrieved at least one header which is the model_cost, that's why i was trying to add it inside the file,
might be from your logic, there is a certain place in responses api that is not acting as the chat completion when retrieving or mapping deployment-name.