scriptoriumadmin/litellm
1
0
Fork 0
mirror of https://github.com/BerriAI/litellm.git synced 2026-09-26 01:12:21 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 98

Merge branch 'BerriAI:litellm_internal_staging' into fix_qdrant_semantic_cache

Browse source
This commit is contained in:
Boris Antonio Duin 2026-05-15 17:55:25 -06:00 • committed by GitHub
commit a18fec166b
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
91 changed files with 969 additions and 818 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

10
litellm/proxy/example_config_yaml/oai_misc_config.yaml
View file

@ -1,7 +1,7 @@
model_list:
- model_name: gpt-3.5-turbo-end-user-test
- model_name: gpt-5-mini-end-user-test
litellm_params:
model: gpt-3.5-turbo
model: gpt-5-mini
region_name: "eu"
model_info:
id: "1"
@ -18,9 +18,9 @@ model_list:
litellm_params:
model: "groq/*"
api_key: os.environ/GROQ_API_KEY
- model_name: bedrock/batch-anthropic.claude-3-5-sonnet-20240620-v1:0
- model_name: bedrock/batch-us.anthropic.claude-haiku-4-5-20251001-v1:0
litellm_params:
model: bedrock/us.anthropic.claude-3-5-sonnet-20240620-v1:0
model: bedrock/us.anthropic.claude-haiku-4-5-20251001-v1:0
#########################################################
########## batch specific params ########################
s3_bucket_name: litellm-proxy
@ -39,7 +39,7 @@ litellm_settings:
num_retries: 5
request_timeout: 600
telemetry: False
context_window_fallbacks: [{"gpt-3.5-turbo": ["gpt-3.5-turbo-large"]}]
context_window_fallbacks: [{"gpt-5-mini": ["gpt-5.5"]}]
default_team_settings:
- team_id: team-1
success_callback: ["langfuse"]

4
litellm/proxy/example_config_yaml/otel_test_config.yaml
View file

@ -1,7 +1,7 @@
model_list:
- model_name: fake-openai-endpoint
litellm_params:
model: openai/gpt-3.5-turbo
model: openai/gpt-5-mini
api_key: fake-key
api_base: https://exampleopenaiendpoint-production.up.railway.app/
tags: ["teamA"]
@ -9,7 +9,7 @@ model_list:
id: "team-a-model"
- model_name: fake-openai-endpoint
litellm_params:
model: openai/gpt-3.5-turbo
model: openai/gpt-5-mini
api_key: fake-key
api_base: https://exampleopenaiendpoint-production.up.railway.app/
tags: ["teamB"]

14
litellm/proxy/example_config_yaml/pass_through_config.yaml
View file

@ -4,21 +4,21 @@ model_list:
model: openai/fake
api_key: fake-key
api_base: https://exampleopenaiendpoint-production.up.railway.app/
- model_name: claude-3-5-sonnet-20241022
- model_name: claude-sonnet-4-5-20250929
litellm_params:
model: anthropic/claude-3-5-sonnet-20241022
model: anthropic/claude-sonnet-4-5-20250929
api_key: os.environ/ANTHROPIC_API_KEY
- model_name: claude-special-alias
litellm_params:
model: anthropic/claude-3-haiku-20240307
model: anthropic/claude-haiku-4-5-20251001
api_key: os.environ/ANTHROPIC_API_KEY
- model_name: claude-3-5-sonnet-20241022
- model_name: claude-sonnet-4-5-20250929
litellm_params:
model: anthropic/claude-3-5-sonnet-20241022
model: anthropic/claude-sonnet-4-5-20250929
api_key: os.environ/ANTHROPIC_API_KEY
- model_name: claude-3-7-sonnet-20250219
- model_name: claude-sonnet-4-6
litellm_params:
model: anthropic/claude-3-7-sonnet-20250219
model: anthropic/claude-sonnet-4-6
api_key: os.environ/ANTHROPIC_API_KEY
- model_name: anthropic/*
litellm_params:

4
litellm/proxy/example_config_yaml/simple_config.yaml
View file

@ -1,4 +1,4 @@
model_list:
- model_name: gpt-3.5-turbo
- model_name: gpt-5-mini
litellm_params:
model: gpt-3.5-turbo
model: gpt-5-mini

2
litellm/proxy/example_config_yaml/spend_tracking_config.yaml
View file

@ -1,7 +1,7 @@
model_list:
- model_name: fake-openai-endpoint
litellm_params:
model: openai/gpt-3.5-turbo
model: openai/gpt-5-mini
api_key: fake-key
api_base: https://exampleopenaiendpoint-production.up.railway.app/

54
litellm/proxy/proxy_server.py
View file

@ -3415,20 +3415,18 @@ class ProxyConfig:
# Make a copy to avoid mutating the original config
config_to_save = new_config.copy()
# SECURITY: Always encrypt environment_variables before DB write
# SECURITY: Always encrypt environment_variables before DB write.
# _encrypt_env_variables_for_db is idempotent — a caller that
# already encrypted the values (or re-submitted ciphertext read
# back from the DB) will not get a stacked second layer.
if (
"environment_variables" in config_to_save
and config_to_save["environment_variables"]
):
# decrypt the environment_variables - in case a caller function has already encrypted the environment_variables
decrypted_env_vars = self._decrypt_and_set_db_env_variables(
environment_variables=config_to_save["environment_variables"],
return_original_value=True,
)
# encrypt the environment_variables,
config_to_save["environment_variables"] = self._encrypt_env_variables(
environment_variables=decrypted_env_vars
config_to_save["environment_variables"] = (
self._encrypt_env_variables_for_db(
environment_variables=config_to_save["environment_variables"]
)
)
config_to_save.pop("model_list", None)
@ -5076,6 +5074,29 @@ class ProxyConfig:
decrypted_variables[k] = decrypted_value
return decrypted_variables
def _encrypt_env_variables_for_db(
self, environment_variables: dict, new_encryption_key: Optional[str] = None
) -> dict:
"""
Idempotently encrypt environment variables for a DB write.
Config writers may pass either plaintext (first write) or values that
are already ciphertext — e.g. the Admin UI reads config back via
/get/config/callbacks (which returns the stored, still-encrypted
value) and re-POSTs it on the next save. Decrypt first so an
already-encrypted value is not stacked with a second encryption
layer, then encrypt exactly once.
Decryption here deliberately uses _decrypt_db_variables (not
_decrypt_and_set_db_env_variables): this is a write path, and
loading values into os.environ is the read path's responsibility.
"""
decrypted_env_vars = self._decrypt_db_variables(environment_variables)
return self._encrypt_env_variables(
environment_variables=decrypted_env_vars,
new_encryption_key=new_encryption_key,
)
@staticmethod
def _parse_router_settings_value(value: Any) -> Optional[dict]:
"""
@ -13788,11 +13809,18 @@ async def update_config( # noqa: PLR0915
existing[k] = v
await _upsert_section("general_settings", existing)
# environment_variables: encrypt request values, then merge into existing.
# environment_variables: idempotently encrypt the request values
# (plaintext on first write, OR ciphertext the UI read back via
# /get/config/callbacks and re-submitted on save), then merge into
# existing. Only the sent keys are re-written; untouched keys keep
# their stored ciphertext byte-for-byte.
if config_info.environment_variables is not None:
existing = await _read_section("environment_variables")
for k, v in config_info.environment_variables.items():
existing[k] = encrypt_value_helper(value=v)
existing.update(
proxy_config._encrypt_env_variables_for_db(
environment_variables=config_info.environment_variables
)
)
await _upsert_section("environment_variables", existing)
# litellm_settings: merge existing + request, request wins (matching

34
proxy_server_config.yaml
View file

@ -1,28 +1,28 @@
model_list:
- model_name: gpt-3.5-turbo-end-user-test
- model_name: gpt-5-mini-end-user-test
litellm_params:
model: gpt-3.5-turbo
model: gpt-5-mini
region_name: "eu"
model_info:
id: "1"
- model_name: gpt-3.5-turbo-end-user-test
- model_name: gpt-5-mini-end-user-test
litellm_params:
model: openai/gpt-4.1-mini
model: openai/gpt-5-mini
api_key: os.environ/OPENAI_API_KEY # The `os.environ/` prefix tells litellm to read this from the env. See https://docs.litellm.ai/docs/simple_proxy#load-api-keys-from-vault
- model_name: gpt-3.5-turbo
litellm_params:
model: openai/gpt-4.1-mini
api_key: os.environ/OPENAI_API_KEY # The `os.environ/` prefix tells litellm to read this from the env. See https://docs.litellm.ai/docs/simple_proxy#load-api-keys-from-vault
- model_name: gpt-3.5-turbo-large
litellm_params:
model: "gpt-3.5-turbo-1106"
litellm_params:
model: "gpt-4.1"
api_key: os.environ/OPENAI_API_KEY
rpm: 480
timeout: 300
stream_timeout: 60
- model_name: gpt-4
litellm_params:
model: openai/gpt-4.1-mini
model: openai/gpt-4.1
api_key: os.environ/OPENAI_API_KEY # The `os.environ/` prefix tells litellm to read this from the env. See https://docs.litellm.ai/docs/simple_proxy#load-api-keys-from-vault
rpm: 480
timeout: 300
@ -32,21 +32,21 @@ model_list:
model: sagemaker/berri-benchmarking-Llama-2-70b-chat-hf-4
input_cost_per_second: 0.000420
- model_name: text-embedding-ada-002
litellm_params:
model: openai/text-embedding-ada-002
litellm_params:
model: openai/text-embedding-3-small
api_key: os.environ/OPENAI_API_KEY
model_info:
mode: embedding
base_model: text-embedding-ada-002
base_model: text-embedding-3-small
- model_name: dall-e-2 # dall-e-2 and dall-e-3 were deprecated 2026-05-12; alias to gpt-image-1
litellm_params:
model: openai/gpt-image-1
- model_name: openai-dall-e-3
- model_name: openai-dall-e-3 # dall-e-3 deprecated 2026-05-12; underlying now gpt-image-1
litellm_params:
model: dall-e-3
model: gpt-image-1
- model_name: fake-openai-endpoint
litellm_params:
model: openai/gpt-3.5-turbo
model: openai/gpt-5-mini
api_key: fake-key
api_base: https://exampleopenaiendpoint-production.up.railway.app/
- model_name: fake-openai-endpoint-2
@ -139,13 +139,13 @@ model_list:
model: openai/my-fake-model
api_key: my-fake-key
api_base: https://exampleopenaiendpoint-production.up.railway.appxxxx/
- model_name: gemini-1.5-flash
- model_name: gemini-2.5-flash
litellm_params:
model: gemini/gemini-1.5-flash
model: gemini/gemini-2.5-flash
api_key: os.environ/GOOGLE_API_KEY
- model_name: gpt-4o
- model_name: gpt-5.5
litellm_params:
model: gpt-4o
model: gpt-5.5
api_key: os.environ/OPENAI_API_KEY

14
tests/batches_tests/test_batches_logging_unit_tests.py
View file

@ -215,7 +215,7 @@ async def test_batch_retrieve_cost_tracking_with_completed_batch_no_explicit_cos
# Create logging object
logging_obj = Logging(
model="gpt-4o-mini",
model="gpt-5-mini",
messages=[{"role": "user", "content": "test"}],
stream=False,
call_type=CallTypes.aretrieve_batch.value,
@ -233,7 +233,7 @@ async def test_batch_retrieve_cost_tracking_with_completed_batch_no_explicit_cos
completion_tokens=50,
total_tokens=150,
)
expected_models = ["gpt-4o-mini"]
expected_models = ["gpt-5-mini"]
with patch(
"litellm.litellm_core_utils.litellm_logging._handle_completed_batch",
@ -299,7 +299,7 @@ async def test_batch_retrieve_cost_tracking_with_explicit_cost_data():
# Create logging object
logging_obj = Logging(
model="gpt-4o-mini",
model="gpt-5-mini",
messages=[{"role": "user", "content": "test"}],
stream=False,
call_type=CallTypes.aretrieve_batch.value,
@ -317,7 +317,7 @@ async def test_batch_retrieve_cost_tracking_with_explicit_cost_data():
completion_tokens=100,
total_tokens=300,
)
explicit_models = ["gpt-4o-mini", "gpt-3.5-turbo"]
explicit_models = ["gpt-5-mini", "gpt-5.5"]
with patch(
"litellm.litellm_core_utils.litellm_logging._handle_completed_batch",
@ -393,7 +393,7 @@ async def test_batch_retrieve_cost_tracking_with_unified_file_id_incomplete_batc
# Create logging object
logging_obj = Logging(
model="gpt-4o-mini",
model="gpt-5-mini",
messages=[{"role": "user", "content": "test"}],
stream=False,
call_type=CallTypes.aretrieve_batch.value,
@ -468,7 +468,7 @@ async def test_batch_retrieve_cost_tracking_with_partial_explicit_data():
# Create logging object
logging_obj = Logging(
model="gpt-4o-mini",
model="gpt-5-mini",
messages=[{"role": "user", "content": "test"}],
stream=False,
call_type=CallTypes.aretrieve_batch.value,
@ -489,7 +489,7 @@ async def test_batch_retrieve_cost_tracking_with_partial_explicit_data():
completion_tokens=75,
total_tokens=225,
)
expected_models = ["gpt-4o-mini"]
expected_models = ["gpt-5-mini"]
with patch(
"litellm.litellm_core_utils.litellm_logging._handle_completed_batch",

104
tests/enterprise/litellm_enterprise/enterprise_callbacks/test_prometheus_logging_callbacks.py
View file

@ -58,9 +58,9 @@ def create_standard_logging_payload() -> StandardLoggingPayload:
endTime=1234567891.0,
completionStartTime=1234567890.5,
model_map_information=StandardLoggingModelInformation(
model_map_key="gpt-3.5-turbo", model_map_value=None
model_map_key="gpt-5-mini", model_map_value=None
),
model="gpt-3.5-turbo",
model="gpt-5-mini",
model_id="model-123",
model_group="openai-gpt",
custom_llm_provider="openai",
@ -109,7 +109,7 @@ def test_safe_get_remaining_budget(prometheus_logger):
async def test_async_log_success_event(prometheus_logger):
standard_logging_object = create_standard_logging_payload()
kwargs = {
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"stream": True,
"litellm_params": {
"metadata": {
@ -208,7 +208,7 @@ def test_increment_token_metrics(prometheus_logger):
end_user_id="user1",
user_api_key="key1",
user_api_key_alias="alias1",
model="gpt-3.5-turbo",
model="gpt-5-mini",
user_api_team="team1",
user_api_team_alias="team_alias1",
user_id="user1",
@ -226,7 +226,7 @@ def test_increment_token_metrics(prometheus_logger):
org_id=None,
org_alias=None,
requested_model=None,
model="gpt-3.5-turbo",
model="gpt-5-mini",
model_id="model-123",
)
prometheus_logger.litellm_tokens_metric.labels().inc.assert_called_once_with(100)
@ -242,7 +242,7 @@ def test_increment_token_metrics(prometheus_logger):
org_id=None,
org_alias=None,
requested_model=None,
model="gpt-3.5-turbo",
model="gpt-5-mini",
model_id="model-123",
)
prometheus_logger.litellm_input_tokens_metric.labels().inc.assert_called_once_with(
@ -260,7 +260,7 @@ def test_increment_token_metrics(prometheus_logger):
org_id=None,
org_alias=None,
requested_model=None,
model="gpt-3.5-turbo",
model="gpt-5-mini",
model_id="model-123",
)
prometheus_logger.litellm_output_tokens_metric.labels().inc.assert_called_once_with(
@ -403,7 +403,7 @@ def test_set_latency_metrics(prometheus_logger):
prometheus_logger._set_latency_metrics(
kwargs=kwargs,
model="gpt-3.5-turbo",
model="gpt-5-mini",
user_api_key="key1",
user_api_key_alias="alias1",
user_api_team="team1",
@ -422,7 +422,7 @@ def test_set_latency_metrics(prometheus_logger):
org_id=None,
org_alias=None,
requested_model="openai-gpt",
model="gpt-3.5-turbo",
model="gpt-5-mini",
model_id="model-123",
)
prometheus_logger.litellm_llm_api_time_to_first_token_metric.labels().observe.assert_called_once_with(
@ -440,7 +440,7 @@ def test_set_latency_metrics(prometheus_logger):
org_id=None,
org_alias=None,
requested_model="openai-gpt",
model="gpt-3.5-turbo",
model="gpt-5-mini",
model_id="model-123",
)
prometheus_logger.litellm_llm_api_latency_metric.labels().observe.assert_called_once_with(
@ -458,7 +458,7 @@ def test_set_latency_metrics(prometheus_logger):
org_id=None,
org_alias=None,
requested_model="openai-gpt",
model="gpt-3.5-turbo",
model="gpt-5-mini",
model_id="model-123",
)
prometheus_logger.litellm_request_total_latency_metric.labels().observe.assert_called_once_with(
@ -497,7 +497,7 @@ def test_set_latency_metrics_missing_timestamps(prometheus_logger):
# This should not raise an exception
prometheus_logger._set_latency_metrics(
kwargs=kwargs,
model="gpt-3.5-turbo",
model="gpt-5-mini",
user_api_key="key1",
user_api_key_alias="alias1",
user_api_team="team1",
@ -544,7 +544,7 @@ def test_set_latency_metrics_missing_api_call_start(prometheus_logger):
# This should not raise an exception
prometheus_logger._set_latency_metrics(
kwargs=kwargs,
model="gpt-3.5-turbo",
model="gpt-5-mini",
user_api_key="key1",
user_api_key_alias="alias1",
user_api_team="team1",
@ -584,7 +584,7 @@ def test_increment_top_level_request_and_spend_metrics(prometheus_logger):
end_user_id="user1",
user_api_key="key1",
user_api_key_alias="alias1",
model="gpt-3.5-turbo",
model="gpt-5-mini",
user_api_team="team1",
user_api_team_alias="team_alias1",
user_id="user1",
@ -602,7 +602,7 @@ def test_increment_top_level_request_and_spend_metrics(prometheus_logger):
team_alias="test_team_alias",
org_id=None,
org_alias=None,
model="gpt-3.5-turbo",
model="gpt-5-mini",
model_id="model-123",
api_provider="openai",
client_ip=None,
@ -621,7 +621,7 @@ def test_increment_top_level_request_and_spend_metrics(prometheus_logger):
team_alias="test_team_alias",
org_id=None,
org_alias=None,
model="gpt-3.5-turbo",
model="gpt-5-mini",
model_id="model-123",
api_provider="openai",
client_ip=None,
@ -635,7 +635,7 @@ async def test_async_log_failure_event(prometheus_logger):
# NOTE: almost all params for this metric are read from standard logging payload
standard_logging_object = create_standard_logging_payload()
kwargs = {
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"litellm_params": {
"custom_llm_provider": "openai",
},
@ -664,7 +664,7 @@ async def test_async_log_failure_event(prometheus_logger):
end_user=None,
hashed_api_key="test_hash",
api_key_alias="test_alias",
model="gpt-3.5-turbo",
model="gpt-5-mini",
team="test_team",
team_alias="test_team_alias",
user="test_user",
@ -674,7 +674,7 @@ async def test_async_log_failure_event(prometheus_logger):
# deployment should be marked in partial outage
prometheus_logger.set_deployment_partial_outage.assert_called_once_with(
litellm_model_name="gpt-3.5-turbo",
litellm_model_name="gpt-5-mini",
model_id="model-123",
api_base="https://api.openai.com",
api_provider="openai",
@ -686,7 +686,7 @@ async def test_async_log_failure_event(prometheus_logger):
prometheus_logger.litellm_deployment_failure_responses.labels.call_args.kwargs
)
expected_failure_labels = {
"litellm_model_name": "gpt-3.5-turbo",
"litellm_model_name": "gpt-5-mini",
"model_id": "model-123",
"api_base": "https://api.openai.com",
"api_provider": "openai",
@ -712,7 +712,7 @@ async def test_async_log_failure_event(prometheus_logger):
prometheus_logger.litellm_deployment_total_requests.labels.call_args.kwargs
)
expected_total_labels = {
"litellm_model_name": "gpt-3.5-turbo",
"litellm_model_name": "gpt-5-mini",
"model_id": "model-123",
"api_base": "https://api.openai.com",
"api_provider": "openai",
@ -788,10 +788,10 @@ async def test_async_post_call_failure_hook(prometheus_logger):
prometheus_logger.litellm_proxy_total_requests_metric = MagicMock()
# Create test data
request_data = {"model": "gpt-3.5-turbo"}
request_data = {"model": "gpt-5-mini"}
original_exception = litellm.RateLimitError(
message="Test error", llm_provider="openai", model="gpt-3.5-turbo"
message="Test error", llm_provider="openai", model="gpt-5-mini"
)
user_api_key_dict = UserAPIKeyAuth(
@ -822,7 +822,7 @@ async def test_async_post_call_failure_hook(prometheus_logger):
team_alias="test_team_alias",
org_id=None,
org_alias=None,
requested_model="gpt-3.5-turbo",
requested_model="gpt-5-mini",
exception_status="429",
exception_class="Openai.RateLimitError",
route=user_api_key_dict.request_route,
@ -837,7 +837,7 @@ async def test_async_post_call_failure_hook(prometheus_logger):
end_user=None,
hashed_api_key="test_key",
api_key_alias="test_alias",
requested_model="gpt-3.5-turbo",
requested_model="gpt-5-mini",
team="test_team",
team_alias="test_team_alias",
org_id=None,
@ -865,7 +865,7 @@ async def test_async_post_call_success_hook(prometheus_logger):
prometheus_logger.litellm_proxy_total_requests_metric = MagicMock()
# Create test data
data = {"model": "gpt-3.5-turbo"}
data = {"model": "gpt-5-mini"}
user_api_key_dict = UserAPIKeyAuth(
api_key="test_key",
@ -909,7 +909,7 @@ def test_set_llm_deployment_success_metrics(prometheus_logger):
# Create test data
request_kwargs = {
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"litellm_params": {
"custom_llm_provider": "openai",
"metadata": {"model_info": {"id": "model-123"}},
@ -946,7 +946,7 @@ def test_set_llm_deployment_success_metrics(prometheus_logger):
model_group="my_custom_model_group", # model_group / requested model from create_standard_logging_payload()
api_provider="openai", # llm provider
api_base="https://api.openai.com", # api base
litellm_model_name="gpt-3.5-turbo", # actual model used - litellm model name
litellm_model_name="gpt-5-mini", # actual model used - litellm model name
hashed_api_key=standard_logging_payload["metadata"]["user_api_key_hash"],
api_key_alias=standard_logging_payload["metadata"]["user_api_key_alias"],
model_id="model-123",
@ -962,7 +962,7 @@ def test_set_llm_deployment_success_metrics(prometheus_logger):
api_key_alias=standard_logging_payload["metadata"]["user_api_key_alias"],
api_provider="openai",
hashed_api_key=standard_logging_payload["metadata"]["user_api_key_hash"],
litellm_model_name="gpt-3.5-turbo",
litellm_model_name="gpt-5-mini",
model_group="my_custom_model_group",
model_id="model-123",
)
@ -973,7 +973,7 @@ def test_set_llm_deployment_success_metrics(prometheus_logger):
# Verify deployment healthy state
prometheus_logger.set_deployment_healthy.assert_called_once_with(
litellm_model_name="gpt-3.5-turbo",
litellm_model_name="gpt-5-mini",
model_id="model-123",
api_base="https://api.openai.com",
api_provider="openai",
@ -981,7 +981,7 @@ def test_set_llm_deployment_success_metrics(prometheus_logger):
# Verify success responses metric
prometheus_logger.litellm_deployment_success_responses.labels.assert_called_once_with(
litellm_model_name="gpt-3.5-turbo",
litellm_model_name="gpt-5-mini",
model_id="model-123",
api_base="https://api.openai.com",
api_provider="openai",
@ -997,7 +997,7 @@ def test_set_llm_deployment_success_metrics(prometheus_logger):
# Verify total requests metric
prometheus_logger.litellm_deployment_total_requests.labels.assert_called_once_with(
litellm_model_name="gpt-3.5-turbo",
litellm_model_name="gpt-5-mini",
model_id="model-123",
api_base="https://api.openai.com",
api_provider="openai",
@ -1013,7 +1013,7 @@ def test_set_llm_deployment_success_metrics(prometheus_logger):
# Verify latency per output token metric
prometheus_logger.litellm_deployment_latency_per_output_token.labels.assert_called_once_with(
litellm_model_name="gpt-3.5-turbo",
litellm_model_name="gpt-5-mini",
model_id="model-123",
api_base="https://api.openai.com",
api_provider="openai",
@ -1029,7 +1029,7 @@ def test_set_llm_deployment_success_metrics(prometheus_logger):
api_key_alias=standard_logging_payload["metadata"]["user_api_key_alias"],
api_provider="openai",
hashed_api_key=standard_logging_payload["metadata"]["user_api_key_hash"],
litellm_model_name="gpt-3.5-turbo",
litellm_model_name="gpt-5-mini",
model_group="my_custom_model_group",
model_id="model-123",
)
@ -1045,9 +1045,9 @@ def test_set_llm_deployment_success_metrics(prometheus_logger):
async def test_log_success_fallback_event(prometheus_logger):
prometheus_logger.litellm_deployment_successful_fallbacks = MagicMock()
original_model_group = "gpt-3.5-turbo"
original_model_group = "gpt-5-mini"
kwargs = {
"model": "gpt-4",
"model": "gpt-5.5",
"metadata": {
"user_api_key_hash": "test_hash",
"user_api_key_alias": "test_alias",
@ -1056,7 +1056,7 @@ async def test_log_success_fallback_event(prometheus_logger):
},
}
original_exception = litellm.RateLimitError(
message="Test error", llm_provider="openai", model="gpt-3.5-turbo"
message="Test error", llm_provider="openai", model="gpt-5-mini"
)
await prometheus_logger.log_success_fallback_event(
@ -1067,7 +1067,7 @@ async def test_log_success_fallback_event(prometheus_logger):
prometheus_logger.litellm_deployment_successful_fallbacks.labels.assert_called_once_with(
requested_model=original_model_group,
fallback_model="gpt-4",
fallback_model="gpt-5.5",
hashed_api_key="test_hash",
api_key_alias="test_alias",
team="test_team",
@ -1083,9 +1083,9 @@ async def test_log_success_fallback_event(prometheus_logger):
async def test_log_failure_fallback_event(prometheus_logger):
prometheus_logger.litellm_deployment_failed_fallbacks = MagicMock()
original_model_group = "gpt-3.5-turbo"
original_model_group = "gpt-5-mini"
kwargs = {
"model": "gpt-4",
"model": "gpt-5.5",
"metadata": {
"user_api_key_hash": "test_hash",
"user_api_key_alias": "test_alias",
@ -1094,7 +1094,7 @@ async def test_log_failure_fallback_event(prometheus_logger):
},
}
original_exception = litellm.RateLimitError(
message="Test error", llm_provider="openai", model="gpt-3.5-turbo"
message="Test error", llm_provider="openai", model="gpt-5-mini"
)
await prometheus_logger.log_failure_fallback_event(
@ -1105,7 +1105,7 @@ async def test_log_failure_fallback_event(prometheus_logger):
prometheus_logger.litellm_deployment_failed_fallbacks.labels.assert_called_once_with(
requested_model=original_model_group,
fallback_model="gpt-4",
fallback_model="gpt-5.5",
hashed_api_key="test_hash",
api_key_alias="test_alias",
team="test_team",
@ -1121,7 +1121,7 @@ def test_deployment_state_management(prometheus_logger):
prometheus_logger.litellm_deployment_state = MagicMock()
test_params = {
"litellm_model_name": "gpt-3.5-turbo",
"litellm_model_name": "gpt-5-mini",
"model_id": "model-123",
"api_base": "https://api.openai.com",
"api_provider": "openai",
@ -1169,7 +1169,7 @@ def test_increment_deployment_cooled_down(prometheus_logger):
)
prometheus_logger.increment_deployment_cooled_down(
litellm_model_name="gpt-3.5-turbo",
litellm_model_name="gpt-5-mini",
model_id="model-123",
api_base="https://api.openai.com",
api_provider="openai",
@ -1177,7 +1177,7 @@ def test_increment_deployment_cooled_down(prometheus_logger):
)
prometheus_logger.litellm_deployment_cooled_down.labels.assert_called_once_with(
"gpt-3.5-turbo", "model-123", "https://api.openai.com", "openai", "429"
"gpt-5-mini", "model-123", "https://api.openai.com", "openai", "429"
)
mock_chain.inc.assert_called_once()
@ -1303,7 +1303,7 @@ async def test_async_log_success_event_with_top_level_metadata(
] = {} # Empty nested dict
kwargs = {
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"stream": True,
"litellm_params": {
"metadata": {
@ -2081,7 +2081,7 @@ def test_get_exception_class_name(prometheus_logger):
"""
# Test case 1: Exception with llm_provider
rate_limit_error = litellm.RateLimitError(
message="Rate limit exceeded", llm_provider="openai", model="gpt-3.5-turbo"
message="Rate limit exceeded", llm_provider="openai", model="gpt-5-mini"
)
assert (
prometheus_logger._get_exception_class_name(rate_limit_error)
@ -2090,7 +2090,7 @@ def test_get_exception_class_name(prometheus_logger):
# Test case 2: Exception with empty llm_provider
auth_error = litellm.AuthenticationError(
message="Invalid API key", llm_provider="", model="gpt-4"
message="Invalid API key", llm_provider="", model="gpt-5.5"
)
assert (
prometheus_logger._get_exception_class_name(auth_error) == "AuthenticationError"
@ -2098,7 +2098,7 @@ def test_get_exception_class_name(prometheus_logger):
# Test case 3: Exception with None llm_provider
context_window_error = litellm.ContextWindowExceededError(
message="Context length exceeded", llm_provider=None, model="gpt-4"
message="Context length exceeded", llm_provider=None, model="gpt-5.5"
)
assert (
prometheus_logger._get_exception_class_name(context_window_error)
@ -2159,7 +2159,7 @@ def test_set_llm_deployment_success_metrics_with_label_filtering():
# Create test data
request_kwargs = {
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"litellm_params": {
"custom_llm_provider": "openai",
"metadata": {"model_info": {"id": "model-123"}},
@ -2310,7 +2310,7 @@ async def test_prometheus_token_metrics_with_prometheus_config():
standard_logging_payload["response_cost"] = 0.075
kwargs = {
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"stream": False,
"litellm_params": {
"metadata": {
@ -2357,7 +2357,7 @@ async def test_prometheus_token_metrics_with_prometheus_config():
expected_label_values = {
"api_key_alias": "test_alias",
"hashed_api_key": "test_hash",
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"team": "test_team",
"team_alias": "test_team_alias",
}

26
tests/enterprise/litellm_enterprise/integrations/test_prometheus.py
View file

@ -110,9 +110,9 @@ def test_end_user_not_tracked_for_all_prometheus_metrics():
team="test_team",
team_alias="test_team_alias",
user="test_user",
requested_model="gpt-4",
model="gpt-4",
litellm_model_name="gpt-4",
requested_model="gpt-5.5",
model="gpt-5.5",
litellm_model_name="gpt-5.5",
)
# Get all defined Prometheus metrics that include end_user in their labels
@ -199,7 +199,7 @@ def test_future_metrics_with_end_user_are_filtered():
hashed_api_key="test_key",
api_key_alias="test_alias",
team="test_team",
model="gpt-4",
model="gpt-5.5",
)
# Test the filtering
@ -556,7 +556,7 @@ async def test_request_counter_semantic_validation(mock_prometheus_logger):
# Test data with large token count that should NOT affect request counter
kwargs = {
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"litellm_params": {"metadata": {}},
"start_time": datetime.now() - timedelta(seconds=1),
"end_time": datetime.now(),
@ -566,7 +566,7 @@ async def test_request_counter_semantic_validation(mock_prometheus_logger):
"prompt_tokens": 600,
"completion_tokens": 399,
"response_cost": 0.005,
"model_group": "gpt-3.5-turbo",
"model_group": "gpt-5-mini",
"model_id": "test-model-id",
"api_base": "https://api.openai.com/v1",
"custom_llm_provider": "openai",
@ -605,7 +605,7 @@ async def test_request_counter_semantic_validation(mock_prometheus_logger):
hashed_api_key="test-hash",
api_key_alias="test-alias",
team="test-team",
model="gpt-4",
model="gpt-5.5",
),
response=MagicMock(),
)
@ -643,7 +643,7 @@ async def test_multiple_requests_counter_semantics(mock_prometheus_logger):
for i in range(num_requests):
kwargs = {
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"litellm_params": {"metadata": {}},
"start_time": datetime.now() - timedelta(seconds=1),
"end_time": datetime.now(),
@ -653,7 +653,7 @@ async def test_multiple_requests_counter_semantics(mock_prometheus_logger):
"prompt_tokens": tokens_per_request // 2,
"completion_tokens": tokens_per_request // 2,
"response_cost": 0.001,
"model_group": "gpt-3.5-turbo",
"model_group": "gpt-5-mini",
"model_id": "test-model-id",
"api_base": "https://api.openai.com/v1",
"custom_llm_provider": "openai",
@ -707,7 +707,7 @@ async def test_streaming_request_counter_semantics(mock_prometheus_logger):
from datetime import datetime, timedelta
kwargs = {
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"litellm_params": {"metadata": {}},
"start_time": datetime.now() - timedelta(seconds=1),
"end_time": datetime.now(),
@ -717,7 +717,7 @@ async def test_streaming_request_counter_semantics(mock_prometheus_logger):
"prompt_tokens": 300,
"completion_tokens": 450,
"response_cost": 0.003,
"model_group": "gpt-3.5-turbo",
"model_group": "gpt-5-mini",
"model_id": "test-model-id",
"api_base": "https://api.openai.com/v1",
"custom_llm_provider": "openai",
@ -801,7 +801,7 @@ async def test_spend_counter_semantics(mock_prometheus_logger):
from datetime import datetime, timedelta
kwargs = {
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"litellm_params": {"metadata": {}},
"start_time": datetime.now() - timedelta(seconds=1),
"end_time": datetime.now(),
@ -811,7 +811,7 @@ async def test_spend_counter_semantics(mock_prometheus_logger):
"prompt_tokens": 60,
"completion_tokens": 40,
"response_cost": 0.0015, # This should be used for spend metrics
"model_group": "gpt-3.5-turbo",
"model_group": "gpt-5-mini",
"model_id": "test-model-id",
"api_base": "https://api.openai.com/v1",
"custom_llm_provider": "openai",

28
tests/enterprise/litellm_enterprise/integrations/test_prometheus_unit_tests.py
View file

@ -78,7 +78,7 @@ async def test_async_prometheus_success_logging_with_callbacks(prometheus_logger
@compare_metrics
async def op():
await litellm.acompletion(
model="claude-3-haiku-20240307",
model="claude-haiku-4-5-20251001",
messages=[{"role": "user", "content": "what llm are u"}],
max_tokens=10,
mock_response="hi",
@ -103,9 +103,9 @@ async def test_async_prometheus_budget_logging_with_callbacks(prometheus_logger)
router = litellm.Router(
model_list=[
{
"model_name": "gpt-3.5-turbo",
"model_name": "gpt-5-mini",
"litellm_params": {
"model": "openai/gpt-3.5-turbo",
"model": "openai/gpt-5-mini",
"api_key": "mock-key",
},
}
@ -114,7 +114,7 @@ async def test_async_prometheus_budget_logging_with_callbacks(prometheus_logger)
)
await router.acompletion(
model="gpt-3.5-turbo",
model="gpt-5-mini",
messages=[{"role": "user", "content": "llm?"}],
mock_response="openai",
metadata={
@ -166,7 +166,7 @@ async def test_prometheus_metric_tracking():
router = Router(
model_list=[
{
"model_name": "gpt-3.5-turbo", # openai model name
"model_name": "gpt-5-mini", # openai model name
"litellm_params": { # params for litellm completion/embedding call
"model": "azure/gpt-4.1-mini",
"api_key": os.getenv("AZURE_AI_API_KEY"),
@ -176,9 +176,9 @@ async def test_prometheus_metric_tracking():
"model_info": {"id": "azure-model-id"},
},
{
"model_name": "gpt-3.5-turbo", # openai model name
"model_name": "gpt-5-mini", # openai model name
"litellm_params": {
"model": "openai/gpt-4o-mini",
"model": "openai/gpt-5-mini",
},
"model_info": {"id": "openai-model-id"},
},
@ -192,7 +192,7 @@ async def test_prometheus_metric_tracking():
try:
response = await router.acompletion(
messages=[{"role": "user", "content": "Hello, how are you?"}],
model="openai/gpt-4o-mini",
model="openai/gpt-5-mini",
mock_response="hi",
)
print(response)
@ -252,8 +252,8 @@ async def test_router_cooldown_event_callback():
# Mock Router instance
mock_router = MagicMock()
mock_deployment = {
"litellm_params": {"model": "gpt-3.5-turbo"},
"model_name": "gpt-3.5-turbo",
"litellm_params": {"model": "gpt-5-mini"},
"model_name": "gpt-5-mini",
"model_info": ModelInfo(id="test-model-id"),
}
mock_router.get_deployment.return_value = mock_deployment
@ -288,13 +288,13 @@ async def test_router_cooldown_event_callback():
assert len(prometheus_logger.deployment_cooled_downs) == 1
assert prometheus_logger.deployment_complete_outages[0] == [
"gpt-3.5-turbo",
"gpt-5-mini",
"test-model-id",
"https://api.openai.com",
"openai",
]
assert prometheus_logger.deployment_cooled_downs[0] == [
"gpt-3.5-turbo",
"gpt-5-mini",
"test-model-id",
"https://api.openai.com",
"openai",
@ -312,8 +312,8 @@ async def test_router_cooldown_event_callback_no_prometheus():
# Mock Router instance
mock_router = MagicMock()
mock_deployment = {
"litellm_params": {"model": "gpt-3.5-turbo"},
"model_name": "gpt-3.5-turbo",
"litellm_params": {"model": "gpt-5-mini"},
"model_name": "gpt-5-mini",
"model_info": ModelInfo(id="test-model-id"),
}
mock_router.get_deployment.return_value = mock_deployment

26
tests/enterprise/litellm_enterprise/proxy/hooks/test_managed_files.py
View file

@ -392,13 +392,13 @@ async def test_router_acreate_batch_only_selects_from_file_id_mapping(monkeypatc
router = litellm.Router(
model_list=[
{
"model_name": "gpt-3.5-turbo",
"litellm_params": {"model": "gpt-3.5-turbo"},
"model_name": "gpt-5-mini",
"litellm_params": {"model": "gpt-5-mini"},
"model_info": {"id": "1234"},
},
{
"model_name": "gpt-3.5-turbo",
"litellm_params": {"model": "gpt-3.5-turbo"},
"model_name": "gpt-5-mini",
"litellm_params": {"model": "gpt-5-mini"},
"model_info": {"id": "5678"},
},
],
@ -413,7 +413,7 @@ async def test_router_acreate_batch_only_selects_from_file_id_mapping(monkeypatc
) as mock_acreate_batch:
for _ in range(1000):
await router.acreate_batch(
model="gpt-3.5-turbo",
model="gpt-5-mini",
input_file_id=file_id,
model_file_id_mapping=model_file_id_mapping,
)
@ -463,7 +463,7 @@ async def test_output_file_id_for_batch_retrieve():
"model_id": "12345679",
"response_cost": 0.0,
"additional_headers": {},
"litellm_model_name": "gpt-4o",
"litellm_model_name": "gpt-5.5",
"unified_batch_id": "litellm_proxy;model_id:12345679;llm_batch_id:batch_685c5e5d63988190b85bdb2147ba131d",
}
proxy_managed_files = _PROXY_LiteLLMManagedFiles(
@ -595,10 +595,10 @@ async def test_error_file_id_for_failed_batch():
"litellm_call_id": "test-call-id",
"api_base": "https://api.openai.com",
"model_id": "test-model-id",
"model_name": "gpt-4o",
"model_name": "gpt-5.5",
"response_cost": 0.0,
"additional_headers": {},
"litellm_model_name": "gpt-4o",
"litellm_model_name": "gpt-5.5",
"unified_batch_id": "litellm_proxy;model_id:test-model-id;llm_batch_id:batch_abc123",
}
@ -667,7 +667,7 @@ async def test_async_post_call_success_hook_twice_assert_no_unique_violation():
batch._hidden_params = {
"model_id": "12345679",
"response_cost": 0.0,
"litellm_model_name": "gpt-4o",
"litellm_model_name": "gpt-5.5",
"unified_batch_id": "litellm_proxy;model_id:12345679;llm_batch_id:batch_685c5e5d63988190b85bdb2147ba131d",
}
@ -1265,7 +1265,7 @@ async def test_completion_with_file_access_check():
],
}
],
"model": "gpt-4",
"model": "gpt-5.5",
}
# Should not raise exception
@ -1331,7 +1331,7 @@ async def test_responses_with_file_access_check():
},
}
],
"model": "gpt-4",
"model": "gpt-5.5",
}
# Should not raise exception
@ -1730,7 +1730,7 @@ async def test_list_batches_from_managed_objects_table_target_model_name_filter_
await proxy_managed_files.list_user_batches(
user_api_key_dict=UserAPIKeyAuth(user_id="test-user"),
limit=10,
target_model_names="gpt-4o,gpt-3.5",
target_model_names="gpt-5.5,gpt-3.5",
)
assert str(exc_info.value) == (
@ -1838,7 +1838,7 @@ async def test_return_unified_file_id_includes_expires_at():
create_file_request=create_file_request,
internal_usage_cache=internal_usage_cache,
litellm_parent_otel_span=None,
target_model_names_list=["gpt-4o"],
target_model_names_list=["gpt-5.5"],
)
# Verify expires_at is passed through

66
tests/enterprise/litellm_enterprise/proxy/management_endpoints/test_project_endpoints_prisma.py
View file

@ -107,10 +107,10 @@ async def test_new_project(prisma_client):
description="Test project for unit testing",
team_id=_team_id,
metadata={"use_case_id": "TEST-001", "responsible_ai_id": "RAI-001"},
models=["gpt-4", "gpt-3.5-turbo"],
models=["gpt-5.5", "gpt-5-mini"],
max_budget=100.0,
model_rpm_limit={"gpt-4": 100},
model_tpm_limit={"gpt-4": 1000},
model_rpm_limit={"gpt-5.5": 100},
model_tpm_limit={"gpt-5.5": 1000},
)
response = await new_project(
@ -130,12 +130,12 @@ async def test_new_project(prisma_client):
assert response.project_alias == "test-project"
assert response.description == "Test project for unit testing"
assert response.team_id == _team_id
assert response.models == ["gpt-4", "gpt-3.5-turbo"]
assert response.models == ["gpt-5.5", "gpt-5-mini"]
# model_rpm_limit and model_tpm_limit are stored in metadata
assert response.metadata["use_case_id"] == "TEST-001"
assert response.metadata["responsible_ai_id"] == "RAI-001"
assert response.metadata["model_rpm_limit"] == {"gpt-4": 100}
assert response.metadata["model_tpm_limit"] == {"gpt-4": 1000}
assert response.metadata["model_rpm_limit"] == {"gpt-5.5": 100}
assert response.metadata["model_tpm_limit"] == {"gpt-5.5": 1000}
assert response.litellm_budget_table is not None
assert response.litellm_budget_table.max_budget == 100.0
@ -181,7 +181,7 @@ async def test_update_project(prisma_client):
metadata={
"use_case_id": "TEST-002",
},
models=["gpt-4"],
models=["gpt-5.5"],
max_budget=50.0,
)
@ -207,10 +207,10 @@ async def test_update_project(prisma_client):
"use_case_id": "TEST-002-UPDATED",
"additional_field": "new_value",
},
models=["gpt-4", "gpt-3.5-turbo", "claude-3"],
models=["gpt-5.5", "gpt-5-mini", "claude-3"],
max_budget=200.0,
model_rpm_limit={"gpt-4": 200, "claude-3": 50},
model_tpm_limit={"gpt-4": 2000, "claude-3": 500},
model_rpm_limit={"gpt-5.5": 200, "claude-3": 50},
model_tpm_limit={"gpt-5.5": 2000, "claude-3": 500},
)
update_response = await update_project(
@ -229,16 +229,16 @@ async def test_update_project(prisma_client):
assert update_response.project_id == project_id
assert update_response.project_alias == "test-project-updated"
assert update_response.description == "Updated description"
assert update_response.models == ["gpt-4", "gpt-3.5-turbo", "claude-3"]
assert update_response.models == ["gpt-5.5", "gpt-5-mini", "claude-3"]
# model_rpm_limit and model_tpm_limit are stored in metadata
assert update_response.metadata["use_case_id"] == "TEST-002-UPDATED"
assert update_response.metadata["additional_field"] == "new_value"
assert update_response.metadata["model_rpm_limit"] == {
"gpt-4": 200,
"gpt-5.5": 200,
"claude-3": 50,
}
assert update_response.metadata["model_tpm_limit"] == {
"gpt-4": 2000,
"gpt-5.5": 2000,
"claude-3": 500,
}
assert update_response.litellm_budget_table is not None
@ -282,7 +282,7 @@ async def test_delete_project(prisma_client):
project_data = NewProjectRequest(
project_alias="test-project-delete",
team_id=_team_id,
models=["gpt-4"],
models=["gpt-5.5"],
max_budget=50.0,
)
@ -374,10 +374,10 @@ async def test_project_info(prisma_client):
description="Test project info endpoint",
team_id=_team_id,
metadata={"use_case_id": "TEST-003", "cost_center": "engineering"},
models=["gpt-4", "claude-3"],
models=["gpt-5.5", "claude-3"],
max_budget=150.0,
model_rpm_limit={"gpt-4": 150},
model_tpm_limit={"gpt-4": 1500},
model_rpm_limit={"gpt-5.5": 150},
model_tpm_limit={"gpt-5.5": 1500},
)
create_response = await new_project(
@ -410,12 +410,12 @@ async def test_project_info(prisma_client):
assert info_response.project_alias == "test-project-info"
assert info_response.description == "Test project info endpoint"
assert info_response.team_id == _team_id
assert info_response.models == ["gpt-4", "claude-3"]
assert info_response.models == ["gpt-5.5", "claude-3"]
# model_rpm_limit and model_tpm_limit are stored in metadata
assert info_response.metadata["use_case_id"] == "TEST-003"
assert info_response.metadata["cost_center"] == "engineering"
assert info_response.metadata["model_rpm_limit"] == {"gpt-4": 150}
assert info_response.metadata["model_tpm_limit"] == {"gpt-4": 1500}
assert info_response.metadata["model_rpm_limit"] == {"gpt-5.5": 150}
assert info_response.metadata["model_tpm_limit"] == {"gpt-5.5": 1500}
assert info_response.litellm_budget_table is not None
assert info_response.litellm_budget_table.max_budget == 150.0
@ -439,12 +439,12 @@ def test_check_team_project_limits_models_not_in_team():
team = LiteLLM_TeamTable(
team_id="test-team",
models=["gpt-4", "gpt-3.5-turbo"],
models=["gpt-5.5", "gpt-5-mini"],
)
data = NewProjectRequest(
team_id="test-team",
models=["gpt-4", "claude-3"], # claude-3 not in team
models=["gpt-5.5", "claude-3"], # claude-3 not in team
)
with pytest.raises(Exception) as exc_info:
@ -465,13 +465,13 @@ def test_check_team_project_limits_budget_exceeds_team():
team = LiteLLM_TeamTable(
team_id="test-team",
models=["gpt-4"],
models=["gpt-5.5"],
max_budget=100.0,
)
data = NewProjectRequest(
team_id="test-team",
models=["gpt-4"],
models=["gpt-5.5"],
max_budget=150.0, # exceeds team's 100.0
)
@ -492,13 +492,13 @@ def test_check_team_project_limits_valid_subset():
team = LiteLLM_TeamTable(
team_id="test-team",
models=["gpt-4", "gpt-3.5-turbo", "claude-3"],
models=["gpt-5.5", "gpt-5-mini", "claude-3"],
max_budget=1000.0,
)
data = NewProjectRequest(
team_id="test-team",
models=["gpt-4", "gpt-3.5-turbo"],
models=["gpt-5.5", "gpt-5-mini"],
max_budget=500.0,
)
@ -522,7 +522,7 @@ def test_check_team_project_limits_all_proxy_models():
data = NewProjectRequest(
team_id="test-team",
models=["gpt-4", "claude-3", "anything-goes"],
models=["gpt-5.5", "claude-3", "anything-goes"],
)
# Should not raise - team allows all models
@ -540,13 +540,13 @@ def test_check_team_project_limits_tpm_exceeds_team():
team = LiteLLM_TeamTable(
team_id="test-team",
models=["gpt-4"],
models=["gpt-5.5"],
tpm_limit=10000,
)
data = NewProjectRequest(
team_id="test-team",
models=["gpt-4"],
models=["gpt-5.5"],
tpm_limit=20000, # exceeds team's 10000
)
@ -567,12 +567,12 @@ def test_check_team_project_limits_negative_budget():
team = LiteLLM_TeamTable(
team_id="test-team",
models=["gpt-4"],
models=["gpt-5.5"],
)
data = NewProjectRequest(
team_id="test-team",
models=["gpt-4"],
models=["gpt-5.5"],
max_budget=-10.0,
)
@ -593,12 +593,12 @@ def test_check_team_project_limits_soft_budget_gte_max():
team = LiteLLM_TeamTable(
team_id="test-team",
models=["gpt-4"],
models=["gpt-5.5"],
)
data = NewProjectRequest(
team_id="test-team",
models=["gpt-4"],
models=["gpt-5.5"],
max_budget=100.0,
soft_budget=100.0, # equal to max, should fail
)

10
tests/guardrails_tests/test_akto_guardrails.py
View file

@ -62,7 +62,7 @@ def akto_ingest():
def sample_inputs() -> GenericGuardrailAPIInputs:
return GenericGuardrailAPIInputs(
texts=["Hello, how are you?"],
model="gpt-4",
model="gpt-5.5",
)
@ -200,7 +200,7 @@ def test_build_akto_payload_format(akto_validate, sample_inputs, sample_request_
req_wrapper = json.loads(payload["requestPayload"])
req_body = json.loads(req_wrapper["body"])
assert req_body["model"] == "gpt-4"
assert req_body["model"] == "gpt-5.5"
assert req_body["messages"][0]["content"] == "Hello, how are you?"
tag = json.loads(payload["tag"])
@ -486,7 +486,7 @@ async def test_fail_open_on_unreachable():
side_effect=httpx.ConnectError("Connection refused")
)
inputs = GenericGuardrailAPIInputs(texts=["test"], model="gpt-4")
inputs = GenericGuardrailAPIInputs(texts=["test"], model="gpt-5.5")
result = await g.apply_guardrail(
inputs=inputs, request_data={}, input_type="request"
)
@ -507,7 +507,7 @@ async def test_fail_closed_on_unreachable():
side_effect=httpx.ConnectError("Connection refused")
)
inputs = GenericGuardrailAPIInputs(texts=["test"], model="gpt-4")
inputs = GenericGuardrailAPIInputs(texts=["test"], model="gpt-5.5")
with pytest.raises(HTTPException) as exc_info:
await g.apply_guardrail(inputs=inputs, request_data={}, input_type="request")
assert exc_info.value.status_code == 503
@ -523,7 +523,7 @@ def test_fail_closed_generic_message():
)
with pytest.raises(HTTPException) as exc_info:
g.handle_unreachable(
inputs=GenericGuardrailAPIInputs(texts=["test"], model="gpt-4"),
inputs=GenericGuardrailAPIInputs(texts=["test"], model="gpt-5.5"),
error=Exception("http://internal-host:9090/secret-path"),
)
assert "internal-host" not in exc_info.value.detail

56
tests/guardrails_tests/test_bedrock_guardrails.py
View file

@ -25,7 +25,7 @@ async def test_bedrock_guardrails_pii_masking():
)
request_data = {
"model": "gpt-4o",
"model": "gpt-5.5",
"messages": [
{"role": "user", "content": "Hello, my phone number is +1 412 555 1212"},
{"role": "assistant", "content": "Hello, how can I help you today?"},
@ -65,7 +65,7 @@ async def test_bedrock_guardrails_pii_masking_content_list():
)
request_data = {
"model": "gpt-4o",
"model": "gpt-5.5",
"messages": [
{
"role": "user",
@ -120,7 +120,7 @@ async def test_bedrock_guardrails_block_messages_api():
)
request_data = {
"model": "claude-3-5-sonnet-20240620",
"model": "claude-sonnet-4-5-20250929",
"messages": [
{
"role": "user",
@ -220,7 +220,7 @@ async def test_bedrock_guardrails_with_streaming():
litellm.callbacks.append(guardrail)
request_data = {
"model": "gpt-4o",
"model": "gpt-5.5",
"messages": [{"role": "user", "content": "Hi I like coffee"}],
"stream": True,
"metadata": {"guardrails": ["bedrock-post-guard"]},
@ -264,7 +264,7 @@ async def test_bedrock_guardrails_with_streaming_no_violation():
litellm.callbacks.append(guardrail)
request_data = {
"model": "gpt-4o",
"model": "gpt-5.5",
"messages": [{"role": "user", "content": "hi"}],
"stream": True,
"metadata": {"guardrails": ["bedrock-post-guard"]},
@ -318,7 +318,7 @@ async def test_bedrock_guardrails_streaming_request_body_mock():
)
],
created=1234567890,
model="gpt-4o",
model="gpt-5.5",
object="chat.completion",
)
@ -333,7 +333,7 @@ async def test_bedrock_guardrails_streaming_request_body_mock():
# Test data - simulating request data and assembled response
request_data = {
"model": "gpt-4o",
"model": "gpt-5.5",
"messages": [{"role": "user", "content": "what's the capital of spain?"}],
"stream": True,
"metadata": {"guardrails": ["bedrock-post-guard"]},
@ -396,7 +396,7 @@ async def test_bedrock_guardrail_aws_param_persistence():
) as mock_get_creds:
for i in range(3):
request_data = {
"model": "gpt-4o",
"model": "gpt-5.5",
"messages": [{"role": "user", "content": f"request {i}"}],
"stream": False,
"metadata": {"guardrails": ["bedrock-post-guard"]},
@ -583,7 +583,7 @@ async def test_bedrock_guardrail_masking_with_anonymized_response():
}
request_data = {
"model": "gpt-4o",
"model": "gpt-5.5",
"messages": [
{"role": "user", "content": "Hello, my phone number is +1 412 555 1212"},
],
@ -657,7 +657,7 @@ async def test_bedrock_guardrail_uses_masked_output_without_masking_flags():
}
request_data = {
"model": "gpt-4o",
"model": "gpt-5.5",
"messages": [
{
"role": "user",
@ -747,12 +747,12 @@ async def test_bedrock_guardrail_response_pii_masking_non_streaming():
)
],
created=1234567890,
model="gpt-4o",
model="gpt-5.5",
object="chat.completion",
)
request_data = {
"model": "gpt-4o",
"model": "gpt-5.5",
"messages": [
{"role": "user", "content": "What's your credit card and phone number?"},
],
@ -834,7 +834,7 @@ async def test_bedrock_guardrail_response_pii_masking_streaming():
)
],
created=1234567890,
model="gpt-4o",
model="gpt-5.5",
object="chat.completion.chunk",
),
ModelResponseStream(
@ -849,7 +849,7 @@ async def test_bedrock_guardrail_response_pii_masking_streaming():
)
],
created=1234567890,
model="gpt-4o",
model="gpt-5.5",
object="chat.completion.chunk",
),
ModelResponseStream(
@ -862,7 +862,7 @@ async def test_bedrock_guardrail_response_pii_masking_streaming():
)
],
created=1234567890,
model="gpt-4o",
model="gpt-5.5",
object="chat.completion.chunk",
),
]
@ -870,7 +870,7 @@ async def test_bedrock_guardrail_response_pii_masking_streaming():
yield chunk
request_data = {
"model": "gpt-4o",
"model": "gpt-5.5",
"messages": [
{"role": "user", "content": "What's your email and SSN?"},
],
@ -1001,7 +1001,7 @@ async def test_convert_to_bedrock_format_output_source():
),
],
created=1234567890,
model="gpt-4o",
model="gpt-5.5",
object="chat.completion",
)
@ -1055,7 +1055,7 @@ async def test_convert_to_bedrock_format_post_call_streaming_hook():
)
],
created=1234567890,
model="gpt-4o",
model="gpt-5.5",
object="chat.completion.chunk",
),
ModelResponseStream(
@ -1068,7 +1068,7 @@ async def test_convert_to_bedrock_format_post_call_streaming_hook():
)
],
created=1234567890,
model="gpt-4o",
model="gpt-5.5",
object="chat.completion.chunk",
),
]
@ -1097,7 +1097,7 @@ async def test_convert_to_bedrock_format_post_call_streaming_hook():
}
request_data = {
"model": "gpt-4o",
"model": "gpt-5.5",
"messages": [{"role": "user", "content": "What's your email?"}],
"stream": True,
}
@ -1223,7 +1223,7 @@ async def test_bedrock_guardrail_blocked_action_shows_output_text():
}
request_data = {
"model": "gpt-4o",
"model": "gpt-5.5",
"messages": [
{"role": "user", "content": "Tell me how to make explosives"},
],
@ -1294,7 +1294,7 @@ async def test_bedrock_guardrail_blocked_action_empty_outputs():
}
request_data = {
"model": "gpt-4o",
"model": "gpt-5.5",
"messages": [
{"role": "user", "content": "Violent content here"},
],
@ -1362,7 +1362,7 @@ async def test_bedrock_guardrail_disable_exception_on_block_non_streaming():
}
request_data = {
"model": "gpt-4o",
"model": "gpt-5.5",
"messages": [
{"role": "user", "content": "Tell me how to make explosives"},
],
@ -1442,7 +1442,7 @@ async def test_bedrock_guardrail_disable_exception_on_block_streaming():
)
],
created=1234567890,
model="gpt-4o",
model="gpt-5.5",
object="chat.completion.chunk",
),
ModelResponseStream(
@ -1455,7 +1455,7 @@ async def test_bedrock_guardrail_disable_exception_on_block_streaming():
)
],
created=1234567890,
model="gpt-4o",
model="gpt-5.5",
object="chat.completion.chunk",
),
]
@ -1480,7 +1480,7 @@ async def test_bedrock_guardrail_disable_exception_on_block_streaming():
}
request_data = {
"model": "gpt-4o",
"model": "gpt-5.5",
"messages": [{"role": "user", "content": "Tell me how to make explosives"}],
"stream": True,
}
@ -1590,12 +1590,12 @@ async def test_bedrock_guardrail_post_call_success_hook_no_output_text():
)
],
created=1234567890,
model="gpt-4o",
model="gpt-5.5",
object="chat.completion",
)
data = {
"model": "gpt-4o",
"model": "gpt-5.5",
"messages": [
{"role": "user", "content": "Hello"},
],

4
tests/guardrails_tests/test_dynamoai_guardrails.py
View file

@ -53,7 +53,7 @@ async def test_dynamoai_blocks_content_with_block_action():
guardrail.async_handler, "post", AsyncMock(return_value=mock_response)
):
request_data = {
"model": "gpt-4",
"model": "gpt-5.5",
"messages": [{"role": "user", "content": "This is harmful content"}],
}
@ -102,7 +102,7 @@ async def test_dynamoai_allows_content_with_none_action():
guardrail.async_handler, "post", AsyncMock(return_value=mock_response)
):
request_data = {
"model": "gpt-4",
"model": "gpt-5.5",
"messages": [{"role": "user", "content": "Hello, how are you?"}],
}

4
tests/guardrails_tests/test_guardrail_load_balancing.py
View file

@ -65,8 +65,8 @@ async def test_proxy_logging_pre_call_hook_load_balancing():
router = Router(
model_list=[
{
"model_name": "gpt-4",
"litellm_params": {"model": "gpt-4", "api_key": "fake-key"},
"model_name": "gpt-5.5",
"litellm_params": {"model": "gpt-5.5", "api_key": "fake-key"},
}
],
guardrail_list=guardrail_list,

4
tests/guardrails_tests/test_guardrails_config.py
View file

@ -58,7 +58,7 @@ def test_guardrail_masking_logging_only():
litellm.callbacks = [callback]
messages = [{"role": "user", "content": "Hey, my name is Peter."}]
response = completion(
model="gpt-3.5-turbo", messages=messages, mock_response="Hi Peter!"
model="gpt-5-mini", messages=messages, mock_response="Hi Peter!"
)
assert response.choices[0].message.content == "Hi Peter!" # type: ignore
@ -82,7 +82,7 @@ def test_guardrail_list_of_event_hooks():
guardrail_name="custom-guard", event_hook=["pre_call", "post_call"]
)
data = {"model": "gpt-3.5-turbo", "metadata": {"guardrails": ["custom-guard"]}}
data = {"model": "gpt-5-mini", "metadata": {"guardrails": ["custom-guard"]}}
assert cg.should_run_guardrail(data=data, event_type=GuardrailEventHooks.pre_call)
assert cg.should_run_guardrail(data=data, event_type=GuardrailEventHooks.post_call)

20
tests/guardrails_tests/test_lakera_v2.py
View file

@ -63,7 +63,7 @@ async def test_lakera_pre_call_hook_for_pii_masking():
"content": "My credit card is 4111-1111-1111-1111 and my email is test@example.com. My phone number is 555-123-4567",
},
],
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"metadata": {},
}
@ -170,7 +170,7 @@ async def test_lakera_blocks_non_pii_violations():
"content": "Some harmful content that triggers violations",
}
],
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"metadata": {},
}
@ -236,7 +236,7 @@ async def test_lakera_only_pii_violations_are_masked():
data = {
"messages": [{"role": "user", "content": "My email test@example.com here"}],
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"metadata": {},
}
@ -423,7 +423,7 @@ async def test_lakera_blocks_flagged_content_with_user_scenario():
"content": "Some harmful content that should be blocked",
}
],
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"metadata": {},
}
@ -487,7 +487,7 @@ async def test_lakera_monitor_mode_allows_flagged_content():
data = {
"messages": [{"role": "user", "content": "Some harmful content"}],
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"metadata": {},
}
@ -535,7 +535,7 @@ async def test_lakera_block_mode_raises_exception():
data = {
"messages": [{"role": "user", "content": "Harmful content"}],
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"metadata": {},
}
@ -578,7 +578,7 @@ async def test_lakera_monitor_mode_during_call():
data = {
"messages": [{"role": "user", "content": "Test content"}],
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"metadata": {},
}
@ -623,7 +623,7 @@ async def test_lakera_post_call_blocks_flagged_content():
data = {
"messages": [{"role": "user", "content": "Harmful content"}],
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"metadata": {},
}
@ -663,7 +663,7 @@ async def test_lakera_post_call_allows_clean_content():
data = {
"messages": [{"role": "user", "content": "Hello"}],
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"metadata": {},
}
@ -713,7 +713,7 @@ async def test_lakera_post_call_masks_pii_and_allows():
data = {
"messages": [{"role": "user", "content": "Hello"}],
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"metadata": {},
}

6
tests/guardrails_tests/test_presidio_pii.py
View file

@ -153,7 +153,7 @@ async def test_presidio_pre_call_hook_with_blocked_entities():
"content": "My credit card is 4111-1111-1111-1111 and my email is test@example.com.",
},
],
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
}
# Mock objects needed for the pre-call hook
@ -201,7 +201,7 @@ async def test_presidio_pre_call_hook_with_different_call_types(call_type):
"content": "My credit card is 4111-1111-1111-1111 and my email is test@example.com. My phone number is 555-123-4567",
},
],
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
}
# Mock objects needed for the pre-call hook
@ -286,7 +286,7 @@ async def test_output_parsing():
]
response = mock_completion(
model="gpt-3.5-turbo",
model="gpt-5-mini",
messages=filtered_message,
mock_response="Hello <PERSON>! How can I assist you today?",
)

14
tests/guardrails_tests/test_tracing_guardrails.py
View file

@ -122,7 +122,7 @@ async def test_standard_logging_payload_includes_guardrail_information():
# 1. call the pre call hook with guardrail
request_data = {
"model": "gpt-4o",
"model": "gpt-5.5",
"messages": [
{"role": "user", "content": "Hello, my phone number is +1 412 555 1212"},
],
@ -221,7 +221,7 @@ async def test_langfuse_trace_includes_guardrail_information():
)
# 1. call the pre call hook with guardrail
request_data = {
"model": "gpt-4o",
"model": "gpt-5.5",
"messages": [
{
"role": "user",
@ -343,7 +343,7 @@ async def test_bedrock_guardrail_status_blocked():
bedrock_guard.async_handler, "post", AsyncMock(return_value=mock_response)
):
request_data = {
"model": "gpt-4o",
"model": "gpt-5.5",
"messages": [{"role": "user", "content": "harmful content"}],
"mock_response": "Hello",
"metadata": {},
@ -440,7 +440,7 @@ async def test_bedrock_guardrail_status_success():
bedrock_guard.async_handler, "post", AsyncMock(return_value=mock_response)
):
request_data = {
"model": "gpt-4o",
"model": "gpt-5.5",
"messages": [{"role": "user", "content": "safe content"}],
"mock_response": "Hello",
"metadata": {},
@ -524,7 +524,7 @@ async def test_bedrock_guardrail_status_failure():
AsyncMock(side_effect=httpx.ConnectError("Connection failed")),
):
request_data = {
"model": "gpt-4o",
"model": "gpt-5.5",
"messages": [{"role": "user", "content": "test content"}],
"mock_response": "Hello",
"metadata": {},
@ -615,7 +615,7 @@ async def test_noma_guardrail_status_blocked():
noma_guard.async_handler, "post", AsyncMock(return_value=mock_response)
):
request_data = {
"model": "gpt-4o",
"model": "gpt-5.5",
"messages": [{"role": "user", "content": "harmful content"}],
"mock_response": "Hello",
"metadata": {},
@ -703,7 +703,7 @@ async def test_noma_guardrail_status_success():
noma_guard.async_handler, "post", AsyncMock(return_value=mock_response)
):
request_data = {
"model": "gpt-4o",
"model": "gpt-5.5",
"messages": [{"role": "user", "content": "safe content"}],
"mock_response": "Hello",
"metadata": {},

40
tests/litellm_utils_tests/test_health_check.py
View file

@ -99,7 +99,7 @@ async def test_azure_img_gen_health_check():
for attempt in range(max_retries):
response = await litellm.ahealth_check(
model_params={
"model": "azure/dall-e-3",
"model": "azure/gpt-image-1",
"api_base": os.getenv("AZURE_AI_API_BASE"),
"api_key": os.getenv("AZURE_AI_API_KEY"),
},
@ -256,9 +256,9 @@ def test_update_litellm_params_for_health_check():
from litellm.proxy.health_check import _update_litellm_params_for_health_check
# Test with health_check_model
model_info = {"health_check_model": "gpt-3.5-turbo"}
model_info = {"health_check_model": "gpt-5-mini"}
litellm_params = {
"model": "gpt-4",
"model": "gpt-5.5",
"api_key": "fake_key",
}
@ -266,12 +266,12 @@ def test_update_litellm_params_for_health_check():
assert "messages" in updated_params
assert isinstance(updated_params["messages"], list)
assert updated_params["model"] == "gpt-3.5-turbo"
assert updated_params["model"] == "gpt-5-mini"
# Test without health_check_model
model_info = {}
litellm_params = {
"model": "gpt-4",
"model": "gpt-5.5",
"api_key": "fake_key",
}
@ -279,12 +279,12 @@ def test_update_litellm_params_for_health_check():
assert "messages" in updated_params
assert isinstance(updated_params["messages"], list)
assert updated_params["model"] == "gpt-4"
assert updated_params["model"] == "gpt-5.5"
# Test with health_check_voice for audio_speech mode
model_info = {"mode": "audio_speech", "health_check_voice": "en-US-JennyNeural"}
litellm_params = {
"model": "gpt-4",
"model": "gpt-5.5",
"api_key": "fake_key",
}
updated_params = _update_litellm_params_for_health_check(model_info, litellm_params)
@ -294,7 +294,7 @@ def test_update_litellm_params_for_health_check():
# Test without health_check_voice for audio_speech mode
model_info = {"mode": "audio_speech"}
litellm_params = {
"model": "gpt-4",
"model": "gpt-5.5",
"api_key": "fake_key",
}
updated_params = _update_litellm_params_for_health_check(model_info, litellm_params)
@ -304,7 +304,7 @@ def test_update_litellm_params_for_health_check():
# Test with health_check_voice for non-audio_speech mode
model_info = {"mode": "chat", "health_check_voice": "en-US-JennyNeural"}
litellm_params = {
"model": "gpt-4",
"model": "gpt-5.5",
"api_key": "fake_key",
}
updated_params = _update_litellm_params_for_health_check(model_info, litellm_params)
@ -339,11 +339,11 @@ def test_update_litellm_params_for_health_check():
# Test that non-Bedrock models are not affected by Bedrock-specific logic
litellm_params = {
"model": "openai/gpt-4",
"model": "openai/gpt-5.5",
"api_key": "fake_key",
}
updated_params = _update_litellm_params_for_health_check(model_info, litellm_params)
assert updated_params["model"] == "openai/gpt-4" # Should remain unchanged
assert updated_params["model"] == "openai/gpt-5.5" # Should remain unchanged
# Test ALL cross-region inference profile prefixes (CRIS)
cris_prefixes = ["us.", "eu.", "apac.", "jp.", "au.", "us-gov.", "global."]
@ -458,14 +458,14 @@ async def test_perform_health_check_filters_by_model_id():
# Two deployments with same model_name but different ids
model_list = [
{
"model_name": "gpt-4",
"model_name": "gpt-5.5",
"model_info": {"id": "deployment-id-1"},
"litellm_params": {"model": "gpt-4", "api_key": "fake-key-1"},
"litellm_params": {"model": "gpt-5.5", "api_key": "fake-key-1"},
},
{
"model_name": "gpt-4",
"model_name": "gpt-5.5",
"model_info": {"id": "deployment-id-2"},
"litellm_params": {"model": "gpt-4", "api_key": "fake-key-2"},
"litellm_params": {"model": "gpt-5.5", "api_key": "fake-key-2"},
},
]
@ -474,7 +474,7 @@ async def test_perform_health_check_filters_by_model_id():
async def mock_perform_health_check(m_list, details=True, **kwargs):
captured_list.append(m_list)
return (
[{"model": "gpt-4", "api_key": m_list[0]["litellm_params"]["api_key"]}],
[{"model": "gpt-5.5", "api_key": m_list[0]["litellm_params"]["api_key"]}],
[],
{},
)
@ -549,7 +549,7 @@ async def test_perform_health_check_with_health_check_model():
"litellm_params": {"model": "openai/*", "api_key": "fake-key"},
"model_info": {
"mode": "chat",
"health_check_model": "openai/gpt-4o-mini", # Override model for health check
"health_check_model": "openai/gpt-5-mini", # Override model for health check
},
}
]
@ -568,10 +568,10 @@ async def test_perform_health_check_with_health_check_model():
print("health check calls: ", health_check_calls)
# Verify the health check used the override model
assert health_check_calls[0] == "openai/gpt-4o-mini"
assert health_check_calls[0] == "openai/gpt-5-mini"
# Verify the result still shows the original model
print("healthy endpoints: ", healthy_endpoints)
assert healthy_endpoints[0]["model"] == "openai/gpt-4o-mini"
assert healthy_endpoints[0]["model"] == "openai/gpt-5-mini"
assert len(healthy_endpoints) == 1
assert len(unhealthy_endpoints) == 0
@ -768,7 +768,7 @@ async def test_image_generation_health_check_prompt(monkeypatch):
model_list = [
{
"litellm_params": {"model": "dall-e-3", "api_key": "fake-key"},
"litellm_params": {"model": "gpt-image-1", "api_key": "fake-key"},
"model_info": {
"mode": "image_generation",
},

66
tests/litellm_utils_tests/test_utils.py
View file

@ -385,14 +385,14 @@ def test_get_valid_models_with_custom_llm_provider(custom_llm_provider):
def test_bad_key():
key = "bad-key"
response = check_valid_key(model="gpt-3.5-turbo", api_key=key)
response = check_valid_key(model="gpt-5-mini", api_key=key)
print(response, key)
assert response == False
def test_good_key():
key = os.environ["OPENAI_API_KEY"]
response = check_valid_key(model="gpt-3.5-turbo", api_key=key)
response = check_valid_key(model="gpt-5-mini", api_key=key)
assert response == True
@ -406,7 +406,7 @@ def test_validate_environment_empty_model():
def test_validate_environment_api_key():
response_obj = validate_environment(model="gpt-3.5-turbo", api_key="sk-my-test-key")
response_obj = validate_environment(model="gpt-5-mini", api_key="sk-my-test-key")
assert (
response_obj["keys_in_environment"] is True
), f"Missing keys={response_obj['missing_keys']}"
@ -598,7 +598,7 @@ def test_get_chat_completion_prompt():
from litellm.litellm_core_utils.litellm_logging import Logging
litellm_logging_obj = Logging(
model="gpt-3.5-turbo",
model="gpt-5-mini",
messages=[{"role": "user", "content": "hi"}],
stream=False,
call_type="acompletion",
@ -610,7 +610,7 @@ def test_get_chat_completion_prompt():
updated_message = "hello world"
litellm_logging_obj.get_chat_completion_prompt(
model="gpt-3.5-turbo",
model="gpt-5-mini",
messages=[{"role": "user", "content": updated_message}],
non_default_params={},
prompt_id="1234",
@ -649,7 +649,7 @@ def test_redact_msgs_from_logs():
)
litellm_logging_obj = Logging(
model="gpt-3.5-turbo",
model="gpt-5-mini",
messages=[{"role": "user", "content": "hi"}],
stream=False,
call_type="acompletion",
@ -700,14 +700,14 @@ def test_redact_embedding_response():
]
response_obj = litellm.EmbeddingResponse(
model="text-embedding-ada-002",
model="text-embedding-3-small",
data=original_data,
usage=original_usage,
object="list",
)
litellm_logging_obj = Logging(
model="text-embedding-ada-002",
model="text-embedding-3-small",
messages=[{"role": "user", "content": "test input"}],
stream=False,
call_type="embedding",
@ -724,13 +724,13 @@ def test_redact_embedding_response():
# Assert the original response_obj is NOT modified
assert response_obj.data == original_data
assert response_obj.usage == original_usage
assert response_obj.model == "text-embedding-ada-002"
assert response_obj.model == "text-embedding-3-small"
assert response_obj.object == "list"
# Assert the redacted response preserves critical metadata
assert _redacted_response_obj.usage == original_usage # usage should be preserved
assert (
_redacted_response_obj.model == "text-embedding-ada-002"
_redacted_response_obj.model == "text-embedding-3-small"
) # model should be preserved
assert _redacted_response_obj.object == "list" # object should be preserved
@ -775,7 +775,7 @@ def test_redact_msgs_from_logs_with_dynamic_params():
)
litellm_logging_obj = Logging(
model="gpt-3.5-turbo",
model="gpt-5-mini",
messages=[{"role": "user", "content": "hi"}],
stream=False,
call_type="acompletion",
@ -934,7 +934,7 @@ def test_logging_trace_id(langfuse_trace_id, langfuse_existing_trace_id):
litellm.success_callback = ["langfuse"]
litellm_call_id = "my-unique-call-id"
litellm_logging_obj = Logging(
model="gpt-3.5-turbo",
model="gpt-5-mini",
messages=[{"role": "user", "content": "hi"}],
stream=False,
call_type="acompletion",
@ -951,7 +951,7 @@ def test_logging_trace_id(langfuse_trace_id, langfuse_existing_trace_id):
metadata["existing_trace_id"] = langfuse_existing_trace_id
litellm.completion(
model="gpt-3.5-turbo",
model="gpt-5-mini",
messages=[{"role": "user", "content": "Hey how's it going?"}],
mock_response="Hey!",
litellm_logging_obj=litellm_logging_obj,
@ -1633,7 +1633,7 @@ def test_get_valid_models_openai_proxy(monkeypatch):
"object": "list",
"data": [
{
"id": "gpt-4o",
"id": "gpt-5.5",
"object": "model",
"created": 1686935002,
"owned_by": "organization-owner",
@ -1650,7 +1650,7 @@ def test_get_valid_models_openai_proxy(monkeypatch):
litellm.module_level_client, "get", return_value=mock_response
) as mock_post:
valid_models = get_valid_models(check_provider_endpoint=True)
assert "litellm_proxy/gpt-4o" in valid_models
assert "litellm_proxy/gpt-5.5" in valid_models
def test_get_valid_models_fireworks_ai(monkeypatch):
@ -1807,7 +1807,7 @@ def test_add_custom_logger_callback_to_specific_event_e2e(monkeypatch):
curr_len_failure_callback = len(litellm.failure_callback)
litellm.completion(
model="gpt-4o-mini",
model="gpt-5-mini",
messages=[{"role": "user", "content": "Hello, world!"}],
mock_response="Testing langfuse",
)
@ -1922,7 +1922,7 @@ async def test_add_custom_logger_callback_to_specific_event_with_duplicates(
# Make a completion call
await litellm.acompletion(
model="gpt-4o-mini",
model="gpt-5-mini",
messages=[{"role": "user", "content": "Hello, world!"}],
mock_response="Testing duplicate callbacks",
)
@ -1961,7 +1961,7 @@ async def test_add_custom_logger_callback_to_specific_event_with_duplicates_succ
# Make a completion call
await litellm.acompletion(
model="gpt-4o-mini",
model="gpt-5-mini",
messages=[{"role": "user", "content": "Hello, world!"}],
mock_response="Testing duplicate callbacks",
)
@ -1996,7 +1996,7 @@ async def test_add_custom_logger_callback_to_specific_event_with_duplicates_call
# Make a completion call
await litellm.acompletion(
model="gpt-4o-mini",
model="gpt-5-mini",
messages=[{"role": "user", "content": "Hello, world!"}],
mock_response="Testing duplicate callbacks",
)
@ -2011,7 +2011,7 @@ async def test_add_custom_logger_callback_to_specific_event_with_duplicates_call
for _ in range(10):
await litellm.acompletion(
model="gpt-4o-mini",
model="gpt-5-mini",
messages=[{"role": "user", "content": "Hello, world!"}],
mock_response="Testing duplicate callbacks",
)
@ -2040,7 +2040,7 @@ def test_add_custom_logger_callback_to_specific_event_e2e_failure(monkeypatch):
curr_len_failure_callback = len(litellm.failure_callback)
litellm.completion(
model="gpt-4o-mini",
model="gpt-5-mini",
messages=[{"role": "user", "content": "Hello, world!"}],
mock_response="Testing langfuse",
)
@ -2069,7 +2069,7 @@ async def test_wrapper_kwargs_passthrough():
return await mock_original(**kwargs)
# Test kwargs
test_kwargs = {"base_model": "gpt-4o-mini"}
test_kwargs = {"base_model": "gpt-5-mini"}
# Call decorated function
await test_function(**test_kwargs)
@ -2089,7 +2089,7 @@ async def test_wrapper_kwargs_passthrough():
# get base model
assert (
litellm_logging_obj.model_call_details["litellm_params"]["base_model"]
== "gpt-4o-mini"
== "gpt-5-mini"
)
@ -2327,15 +2327,15 @@ def test_get_valid_models_from_provider():
valid_models = get_valid_models(custom_llm_provider="openai")
assert len(valid_models) > 0
assert "gpt-4o-mini" in valid_models
assert "gpt-5-mini" in valid_models
print("Valid models: ", valid_models)
valid_models.remove("gpt-4o-mini")
assert "gpt-4o-mini" not in valid_models
valid_models.remove("gpt-5-mini")
assert "gpt-5-mini" not in valid_models
valid_models = get_valid_models(custom_llm_provider="openai")
assert len(valid_models) > 0
assert "gpt-4o-mini" in valid_models
assert "gpt-5-mini" in valid_models
def test_get_valid_models_from_provider_cache_invalidation(monkeypatch):
@ -2347,7 +2347,7 @@ def test_get_valid_models_from_provider_cache_invalidation(monkeypatch):
monkeypatch.setenv("OPENAI_API_KEY", "123")
_model_cache.set_cached_model_info(
"openai", litellm_params=None, available_models=["gpt-4o-mini"]
"openai", litellm_params=None, available_models=["gpt-5-mini"]
)
monkeypatch.delenv("OPENAI_API_KEY")
@ -2471,10 +2471,10 @@ def test_get_base_model_from_metadata():
# Test 1: base_model in metadata (Chat Completions API pattern)
model_call_details_with_metadata = {
"litellm_params": {"metadata": {"model_info": {"base_model": "azure/gpt-4"}}}
"litellm_params": {"metadata": {"model_info": {"base_model": "azure/gpt-5.5"}}}
}
result = _get_base_model_from_metadata(model_call_details_with_metadata)
assert result == "azure/gpt-4", f"Expected 'azure/gpt-4', got {result}"
assert result == "azure/gpt-5.5", f"Expected 'azure/gpt-5.5', got {result}"
# Test 2: base_model in litellm_metadata (Responses API and generic API calls pattern)
model_call_details_with_litellm_metadata = {
@ -2487,12 +2487,12 @@ def test_get_base_model_from_metadata():
# Test 3: base_model in litellm_params (direct base_model)
model_call_details_with_direct_base_model = {
"litellm_params": {"base_model": "azure/gpt-3.5-turbo"}
"litellm_params": {"base_model": "azure/gpt-5-mini"}
}
result = _get_base_model_from_metadata(model_call_details_with_direct_base_model)
assert (
result == "azure/gpt-3.5-turbo"
), f"Expected 'azure/gpt-3.5-turbo', got {result}"
result == "azure/gpt-5-mini"
), f"Expected 'azure/gpt-5-mini', got {result}"
# Test 4: metadata takes precedence over litellm_metadata
model_call_details_with_both = {

6
tests/llm_responses_api_testing/base_responses_api.py
View file

@ -363,7 +363,7 @@ class BaseResponsesAPITest(ABC):
litellm._turn_on_debug()
response = await litellm.aresponses(
model="gpt-4o",
model="gpt-5.5",
input="Tell me a three sentence bedtime story about a unicorn.",
)
print("Initial response=", json.dumps(response, indent=4, default=str))
@ -771,7 +771,7 @@ class BaseResponsesAPITest(ABC):
except litellm.BadRequestError as e:
if "shell" in str(e).lower() and "not supported" in str(e).lower():
pytest.skip(
"Shell tool is not supported for this model (e.g. gpt-4o); use a model that supports shell"
"Shell tool is not supported for this model (e.g. gpt-5.5); use a model that supports shell"
)
raise
validate_responses_api_response(response, final_chunk=True)
@ -785,7 +785,7 @@ class BaseResponsesAPITest(ABC):
Calls aresponses(..., tools=[shell], stream=True), then iterates the stream and
asserts at least one event is shell-related or response output contains shell_call.
Skips when model does not support shell (e.g. gpt-4o).
Skips when model does not support shell (e.g. gpt-5.5).
"""
base_completion_call_args = self.get_base_completion_call_args()
model = (

20
tests/llm_responses_api_testing/test_base_responses_api_streaming_iterator.py
View file

@ -72,7 +72,7 @@ class TestBaseResponsesAPIStreamingIterator:
# Create the iterator instance
iterator = BaseResponsesAPIStreamingIterator(
response=mock_response,
model="gpt-4",
model="gpt-5.5",
responses_api_provider_config=mock_config,
logging_obj=mock_logging_obj,
litellm_metadata={"model_info": {"id": "model_123"}},
@ -142,7 +142,7 @@ class TestBaseResponsesAPIStreamingIterator:
# Create the iterator instance
iterator = BaseResponsesAPIStreamingIterator(
response=mock_response,
model="gpt-4",
model="gpt-5.5",
responses_api_provider_config=mock_config,
logging_obj=mock_logging_obj,
litellm_metadata={"model_info": {"id": "model_123"}},
@ -188,7 +188,7 @@ class TestBaseResponsesAPIStreamingIterator:
# Create the iterator instance
iterator = BaseResponsesAPIStreamingIterator(
response=mock_response,
model="gpt-4",
model="gpt-5.5",
responses_api_provider_config=mock_config,
logging_obj=mock_logging_obj,
)
@ -214,7 +214,7 @@ class TestBaseResponsesAPIStreamingIterator:
# Create the iterator instance
iterator = BaseResponsesAPIStreamingIterator(
response=mock_response,
model="gpt-4",
model="gpt-5.5",
responses_api_provider_config=mock_config,
logging_obj=mock_logging_obj,
)
@ -240,7 +240,7 @@ class TestBaseResponsesAPIStreamingIterator:
# Create the iterator instance
iterator = BaseResponsesAPIStreamingIterator(
response=mock_response,
model="gpt-4",
model="gpt-5.5",
responses_api_provider_config=mock_config,
logging_obj=mock_logging_obj,
)
@ -280,7 +280,7 @@ class TestBaseResponsesAPIStreamingIterator:
# Create the iterator instance
iterator = ResponsesAPIStreamingIterator(
response=mock_response,
model="gpt-4",
model="gpt-5.5",
responses_api_provider_config=mock_config,
logging_obj=mock_logging_obj,
litellm_metadata={"model_info": {"id": "model_123"}},
@ -355,7 +355,7 @@ class TestBaseResponsesAPIStreamingIterator:
# Create the iterator instance
iterator = ResponsesAPIStreamingIterator(
response=mock_response,
model="gpt-4",
model="gpt-5.5",
responses_api_provider_config=mock_config,
logging_obj=mock_logging_obj,
litellm_metadata={"model_info": {"id": "model_123"}},
@ -417,7 +417,7 @@ class TestBaseResponsesAPIStreamingIterator:
# Create the iterator instance
iterator = SyncResponsesAPIStreamingIterator(
response=mock_response,
model="gpt-4",
model="gpt-5.5",
responses_api_provider_config=mock_config,
logging_obj=mock_logging_obj,
litellm_metadata={"model_info": {"id": "model_123"}},
@ -475,7 +475,7 @@ class TestBaseResponsesAPIStreamingIterator:
iterator = ResponsesAPIStreamingIterator(
response=mock_response,
model="gpt-4",
model="gpt-5.5",
responses_api_provider_config=mock_config,
logging_obj=mock_logging_obj,
litellm_metadata={"model_info": {"id": "model_123"}},
@ -554,7 +554,7 @@ class TestBaseResponsesAPIStreamingIterator:
iterator = ResponsesAPIStreamingIterator(
response=mock_response,
model="gpt-4",
model="gpt-5.5",
responses_api_provider_config=mock_config,
logging_obj=mock_logging_obj,
litellm_metadata={"model_info": {"id": "model_123"}},

72
tests/llm_responses_api_testing/test_openai_responses_api.py
View file

@ -28,7 +28,7 @@ from base_responses_api import BaseResponsesAPITest, validate_responses_api_resp
class TestOpenAIResponsesAPITest(BaseResponsesAPITest):
def get_base_completion_call_args(self):
return {
"model": "openai/gpt-4o",
"model": "openai/gpt-5.5",
}
def get_base_completion_reasoning_call_args(self):
@ -104,7 +104,7 @@ def test_basic_openai_responses_api_streaming_with_logging():
litellm.set_verbose = True
test_custom_logger = TestCustomLogger()
litellm.callbacks = [test_custom_logger]
request_model = "gpt-4o"
request_model = "gpt-5.5"
response = litellm.responses(
model=request_model,
input="hi",
@ -176,7 +176,7 @@ async def test_basic_openai_responses_api_non_streaming_with_logging():
litellm.set_verbose = True
test_custom_logger = TestCustomLogger()
litellm.callbacks = [test_custom_logger]
request_model = "gpt-4o"
request_model = "gpt-5.5"
response = await litellm.aresponses(
model=request_model,
input="hi",
@ -215,13 +215,13 @@ async def test_openai_responses_api_returns_headers(sync_mode):
if sync_mode:
response = litellm.responses(
model="gpt-4o",
model="gpt-5.5",
input="Say hello",
max_output_tokens=20,
)
else:
response = await litellm.aresponses(
model="gpt-4o",
model="gpt-5.5",
input="Say hello",
max_output_tokens=20,
)
@ -471,7 +471,7 @@ async def test_openai_responses_api_streaming_validation(sync_mode):
if sync_mode:
response = litellm.responses(
model="gpt-4o",
model="gpt-5.5",
input="Tell me about artificial intelligence in 3 sentences.",
stream=True,
)
@ -481,7 +481,7 @@ async def test_openai_responses_api_streaming_validation(sync_mode):
event_types_seen.add(event.type)
else:
response = await litellm.aresponses(
model="gpt-4o",
model="gpt-5.5",
input="Tell me about artificial intelligence in 3 sentences.",
stream=True,
)
@ -511,7 +511,7 @@ async def test_openai_responses_litellm_router(sync_mode):
{
"model_name": "gpt4o-special-alias",
"litellm_params": {
"model": "gpt-4o",
"model": "gpt-5.5",
"api_key": os.getenv("OPENAI_API_KEY"),
},
}
@ -556,7 +556,7 @@ async def test_openai_responses_litellm_router_streaming(sync_mode):
{
"model_name": "gpt4o-special-alias",
"litellm_params": {
"model": "gpt-4o",
"model": "gpt-5.5",
"api_key": os.getenv("OPENAI_API_KEY"),
},
}
@ -605,7 +605,7 @@ async def test_openai_responses_litellm_router_no_metadata():
"object": "response",
"created_at": 1741476542,
"status": "completed",
"model": "gpt-4o",
"model": "gpt-5.5",
"output": [
{
"type": "message",
@ -664,7 +664,7 @@ async def test_openai_responses_litellm_router_no_metadata():
{
"model_name": "gpt4o-special-alias",
"litellm_params": {
"model": "gpt-4o",
"model": "gpt-5.5",
"api_key": "fake-key",
},
}
@ -704,7 +704,7 @@ async def test_openai_responses_litellm_router_with_metadata():
"object": "response",
"created_at": 1741476542,
"status": "completed",
"model": "gpt-4o",
"model": "gpt-5.5",
"output": [
{
"type": "message",
@ -762,7 +762,7 @@ async def test_openai_responses_litellm_router_with_metadata():
{
"model_name": "gpt4o-special-alias",
"litellm_params": {
"model": "gpt-4o",
"model": "gpt-5.5",
"api_key": "fake-key",
},
}
@ -802,7 +802,7 @@ async def test_openai_responses_litellm_router_with_prompt():
"object": "response",
"created_at": 1741476542,
"status": "completed",
"model": "gpt-4o",
"model": "gpt-5.5",
"output": [],
"parallel_tool_calls": True,
"usage": {"input_tokens": 0, "output_tokens": 0, "total_tokens": 0},
@ -844,7 +844,7 @@ async def test_openai_responses_litellm_router_with_prompt():
{
"model_name": "gpt4o-special-alias",
"litellm_params": {
"model": "gpt-4o",
"model": "gpt-5.5",
"api_key": "fake-key",
},
}
@ -865,7 +865,7 @@ async def test_openai_responses_litellm_router_with_prompt():
def test_bad_request_bad_param_error():
"""Raise a BadRequestError when an invalid parameter value is provided"""
try:
litellm.responses(model="gpt-4o", input="This should fail", temperature=2000)
litellm.responses(model="gpt-5.5", input="This should fail", temperature=2000)
pytest.fail("Expected BadRequestError but no exception was raised")
except litellm.BadRequestError as e:
print(f"Exception raised: {e}")
@ -881,7 +881,7 @@ async def test_async_bad_request_bad_param_error():
"""Raise a BadRequestError when an invalid parameter value is provided"""
try:
await litellm.aresponses(
model="gpt-4o", input="This should fail", temperature=2000
model="gpt-5.5", input="This should fail", temperature=2000
)
pytest.fail("Expected BadRequestError but no exception was raised")
except litellm.BadRequestError as e:
@ -1280,7 +1280,7 @@ async def test_openai_responses_api_field_types():
# Test with store=True
response = await litellm.aresponses(
model="gpt-4o",
model="gpt-5.5",
input="hi",
)
@ -1292,7 +1292,7 @@ async def test_openai_responses_api_field_types():
assert response.store is True, "store field should match input value"
# Test without store parameter
response_without_store = await litellm.aresponses(model="gpt-4o", input="hi")
response_without_store = await litellm.aresponses(model="gpt-5.5", input="hi")
# Verify created_at is still an integer
assert isinstance(
@ -1310,7 +1310,7 @@ async def test_store_field_transformation():
# Initialize logging object with required parameters
logging_obj = LiteLLMLoggingObj(
model="gpt-4o",
model="gpt-5.5",
messages=[],
stream=False,
call_type="aresponses",
@ -1323,7 +1323,7 @@ async def test_store_field_transformation():
base_response = {
"id": "test_id",
"created_at": 1751443898,
"model": "gpt-4o",
"model": "gpt-5.5",
"object": "response",
"output": [
{
@ -1378,7 +1378,7 @@ async def test_store_field_transformation():
# Test when store=True in request
logging_obj.optional_params = {"store": True}
response = config.transform_response_api_response(
model="gpt-4o", raw_response=mock_response_store_true, logging_obj=logging_obj
model="gpt-5.5", raw_response=mock_response_store_true, logging_obj=logging_obj
)
assert (
response.store is True
@ -1387,7 +1387,7 @@ async def test_store_field_transformation():
# Test when store=False in request
logging_obj.optional_params = {"store": False}
response = config.transform_response_api_response(
model="gpt-4o", raw_response=mock_response_store_false, logging_obj=logging_obj
model="gpt-5.5", raw_response=mock_response_store_false, logging_obj=logging_obj
)
assert (
response.store is False
@ -1395,7 +1395,7 @@ async def test_store_field_transformation():
# Test when store not in request but API returns null
response = config.transform_response_api_response(
model="gpt-4o", raw_response=mock_response_store_null, logging_obj=logging_obj
model="gpt-5.5", raw_response=mock_response_store_null, logging_obj=logging_obj
)
assert (
response.store is None
@ -1403,7 +1403,7 @@ async def test_store_field_transformation():
# Test when store not in request and API omits store field
response = config.transform_response_api_response(
model="gpt-4o", raw_response=mock_response_no_store, logging_obj=logging_obj
model="gpt-5.5", raw_response=mock_response_no_store, logging_obj=logging_obj
)
assert (
response.store is None
@ -1484,7 +1484,7 @@ async def test_aresponses_service_tier_and_safety_identifier():
# Call aresponses with service_tier and safety_identifier
response = await litellm.aresponses(
model="openai/gpt-4o",
model="openai/gpt-5.5",
input="Test with service tier and safety identifier",
service_tier="flex",
safety_identifier="123",
@ -1502,7 +1502,7 @@ async def test_aresponses_service_tier_and_safety_identifier():
assert (
request_body["safety_identifier"] == "123"
), "safety_identifier should be '123' in request body"
assert request_body["model"] == "gpt-4o"
assert request_body["model"] == "gpt-5.5"
assert request_body["input"] == "Test with service tier and safety identifier"
# Validate the response
@ -1609,7 +1609,7 @@ async def test_openai_gpt5_reasoning_effort_parameter():
@pytest.mark.parametrize("stream", [True, False])
async def test_basic_openai_responses_with_websearch(stream):
litellm._turn_on_debug()
request_model = "gpt-4o"
request_model = "gpt-5.5"
response = await litellm.aresponses(
model=request_model,
stream=stream,
@ -1715,7 +1715,7 @@ def extra_body_mock_response_data():
"object": "response",
"created_at": 1234567890,
"status": "completed",
"model": "gpt-4o",
"model": "gpt-5.5",
"output": [
{
"type": "message",
@ -1747,7 +1747,7 @@ async def test_aresponses_extra_body_params_passed(extra_body_mock_response_data
mock_post.return_value = MockResponse(extra_body_mock_response_data, 200)
response = await litellm.aresponses(
model="gpt-4o",
model="gpt-5.5",
input="Test input",
max_output_tokens=20,
extra_body={
@ -1768,7 +1768,7 @@ async def test_aresponses_extra_body_params_passed(extra_body_mock_response_data
assert request_body["custom_param_2"]["nested"] == "value2"
assert "experimental_feature" in request_body
assert request_body["experimental_feature"] is True
assert request_body["model"] == "gpt-4o"
assert request_body["model"] == "gpt-5.5"
assert request_body["input"] == "Test input"
@ -1779,7 +1779,7 @@ def test_responses_extra_body_params_passed_sync(extra_body_mock_response_data):
return_value=MockResponse(extra_body_mock_response_data, 200),
) as mock_post:
response = litellm.responses(
model="gpt-4o",
model="gpt-5.5",
input="Sync test",
max_output_tokens=20,
extra_body={
@ -1797,7 +1797,7 @@ def test_responses_extra_body_params_passed_sync(extra_body_mock_response_data):
assert request_body["sync_custom_param"] == "sync_value"
assert "another_param" in request_body
assert request_body["another_param"] == 42
assert request_body["model"] == "gpt-4o"
assert request_body["model"] == "gpt-5.5"
@pytest.mark.asyncio
@ -1810,7 +1810,7 @@ async def test_extra_body_merges_with_request_data(extra_body_mock_response_data
mock_post.return_value = MockResponse(extra_body_mock_response_data, 200)
await litellm.aresponses(
model="gpt-4o",
model="gpt-5.5",
input="Test",
temperature=0.7,
max_output_tokens=20,
@ -1847,13 +1847,13 @@ async def test_openai_compact_responses_api(sync_mode):
try:
if sync_mode:
response = litellm.compact_responses(
model="openai/gpt-4o",
model="openai/gpt-5.5",
input=input_messages,
instructions="Be helpful and concise",
)
else:
response = await litellm.acompact_responses(
model="openai/gpt-4o",
model="openai/gpt-5.5",
input=input_messages,
instructions="Be helpful and concise",
)

8
tests/llm_translation/test_anthropic_completion.py
View file

@ -1379,7 +1379,7 @@ def test_anthropic_mcp_server_tool_use(spec: str):
]
params = {
"model": "anthropic/claude-sonnet-4-20250514",
"model": "anthropic/claude-sonnet-4-5-20250929",
"messages": [{"role": "user", "content": "Who won the World Cup in 2022?"}],
"tools": tools,
}
@ -1392,7 +1392,7 @@ def test_anthropic_mcp_server_tool_use(spec: str):
@pytest.mark.parametrize(
"model", ["openai/gpt-4.1", "anthropic/claude-sonnet-4-20250514"]
"model", ["openai/gpt-4.1", "anthropic/claude-sonnet-4-5-20250929"]
)
@pytest.mark.skipif(
os.getenv("ZAPIER_CI_CD_MCP_TOKEN") is None, reason="ZAPIER_CI_CD_MCP_TOKEN not set"
@ -1506,8 +1506,8 @@ def test_anthropic_tool_cache_control():
}
]
vertex_ai_model = "vertex_ai/claude-sonnet-4@20250514"
anthropic_api_model = "claude-sonnet-4-20250514"
vertex_ai_model = "vertex_ai/claude-sonnet-4-5@20250929"
anthropic_api_model = "claude-sonnet-4-5-20250929"
result = return_raw_request(
endpoint=CallTypes.completion,
kwargs={

4
tests/llm_translation/test_optional_params.py
View file

@ -2037,7 +2037,7 @@ def test_drop_store_param_for_anthropic():
Ref: https://github.com/BerriAI/litellm/issues/19700
"""
optional_params = get_optional_params(
model="claude-sonnet-4-20250514",
model="claude-sonnet-4-5-20250929",
custom_llm_provider="anthropic",
drop_params=True,
store=True,
@ -2053,7 +2053,7 @@ def test_additional_drop_params_store_for_anthropic():
Ref: https://github.com/BerriAI/litellm/issues/19700
"""
optional_params = get_optional_params(
model="claude-sonnet-4-20250514",
model="claude-sonnet-4-5-20250929",
custom_llm_provider="anthropic",
additional_drop_params=["store"],
store=True,

8
tests/local_testing/create_mock_standard_logging_payload.py
View file

@ -43,9 +43,9 @@ def create_standard_logging_payload() -> StandardLoggingPayload:
endTime=1234567891.0,
completionStartTime=1234567890.5,
model_map_information=StandardLoggingModelInformation(
model_map_key="gpt-3.5-turbo", model_map_value=None
model_map_key="gpt-5-mini", model_map_value=None
),
model="gpt-3.5-turbo",
model="gpt-5-mini",
model_id="model-123",
model_group="openai-gpt",
api_base="https://api.openai.com",
@ -94,9 +94,9 @@ def create_standard_logging_payload_with_long_content() -> StandardLoggingPayloa
endTime=1234567891.0,
completionStartTime=1234567890.5,
model_map_information=StandardLoggingModelInformation(
model_map_key="gpt-3.5-turbo", model_map_value=None
model_map_key="gpt-5-mini", model_map_value=None
),
model="gpt-3.5-turbo",
model="gpt-5-mini",
model_id="model-123",
model_group="openai-gpt",
api_base="https://api.openai.com",

8
tests/logging_callback_tests/create_mock_standard_logging_payload.py
View file

@ -43,9 +43,9 @@ def create_standard_logging_payload() -> StandardLoggingPayload:
endTime=1234567891.0,
completionStartTime=1234567890.5,
model_map_information=StandardLoggingModelInformation(
model_map_key="gpt-3.5-turbo", model_map_value=None
model_map_key="gpt-5-mini", model_map_value=None
),
model="gpt-3.5-turbo",
model="gpt-5-mini",
model_id="model-123",
model_group="openai-gpt",
api_base="https://api.openai.com",
@ -94,9 +94,9 @@ def create_standard_logging_payload_with_long_content() -> StandardLoggingPayloa
endTime=1234567891.0,
completionStartTime=1234567890.5,
model_map_information=StandardLoggingModelInformation(
model_map_key="gpt-3.5-turbo", model_map_value=None
model_map_key="gpt-5-mini", model_map_value=None
),
model="gpt-3.5-turbo",
model="gpt-5-mini",
model_id="model-123",
model_group="openai-gpt",
api_base="https://api.openai.com",

40
tests/logging_callback_tests/test_alerting.py
View file

@ -43,7 +43,7 @@ from litellm.utils import get_api_base
"model, optional_params, expected_api_base",
[
("openai/my-fake-model", {"api_base": "my-fake-api-base"}, "my-fake-api-base"),
("gpt-3.5-turbo", {}, "https://api.openai.com"),
("gpt-5-mini", {}, "https://api.openai.com"),
],
)
def test_get_api_base_unit_test(model, optional_params, expected_api_base):
@ -254,7 +254,7 @@ async def test_daily_reports_unit_test(slack_alerting):
model_list=[
{
"model_name": "test-gpt",
"litellm_params": {"model": "gpt-3.5-turbo"},
"litellm_params": {"model": "gpt-5-mini"},
"model_info": {"id": "1234"},
}
]
@ -286,16 +286,16 @@ async def test_daily_reports_completion(slack_alerting):
router = litellm.Router(
model_list=[
{
"model_name": "gpt-5",
"model_name": "gpt-5.5",
"litellm_params": {
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
},
}
]
)
await router.acompletion(
model="gpt-3.5-turbo",
model="gpt-5-mini",
messages=[{"role": "user", "content": "Hey, how's it going?"}],
)
@ -310,15 +310,15 @@ async def test_daily_reports_completion(slack_alerting):
router = litellm.Router(
model_list=[
{
"model_name": "gpt-5",
"litellm_params": {"model": "gpt-3.5-turbo", "api_key": "bad_key"},
"model_name": "gpt-5.5",
"litellm_params": {"model": "gpt-5-mini", "api_key": "bad_key"},
}
]
)
try:
await router.acompletion(
model="gpt-3.5-turbo",
model="gpt-5-mini",
messages=[{"role": "user", "content": "Hey, how's it going?"}],
)
except Exception as e:
@ -347,9 +347,9 @@ async def test_daily_reports_redis_cache_scheduler():
router = litellm.Router(
model_list=[
{
"model_name": "gpt-5",
"model_name": "gpt-5.5",
"litellm_params": {
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
},
}
]
@ -388,16 +388,16 @@ async def test_send_llm_exception_to_slack():
router = litellm.Router(
model_list=[
{
"model_name": "gpt-3.5-turbo",
"model_name": "gpt-5-mini",
"litellm_params": {
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"api_key": "bad_key",
},
},
{
"model_name": "gpt-5-good",
"litellm_params": {
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
},
},
],
@ -407,7 +407,7 @@ async def test_send_llm_exception_to_slack():
)
try:
await router.acompletion(
model="gpt-3.5-turbo",
model="gpt-5-mini",
messages=[{"role": "user", "content": "Hey, how's it going?"}],
)
except Exception:
@ -582,9 +582,9 @@ async def test_webhook_alerting(alerting_type):
@pytest.mark.parametrize(
"model, api_base, llm_provider, vertex_project, vertex_location",
[
("gpt-3.5-turbo", None, "openai", None, None),
("gpt-5-mini", None, "openai", None, None),
(
"azure/gpt-3.5-turbo",
"azure/gpt-5-mini",
"https://openai-gpt-4-test-v-1.openai.azure.com",
"azure",
None,
@ -688,9 +688,9 @@ async def test_outage_alerting_called(
@pytest.mark.parametrize(
"model, api_base, llm_provider, vertex_project, vertex_location",
[
("gpt-3.5-turbo", None, "openai", None, None),
("gpt-5-mini", None, "openai", None, None),
(
"azure/gpt-3.5-turbo",
"azure/gpt-5-mini",
"https://openai-gpt-4-test-v-1.openai.azure.com",
"azure",
None,
@ -800,7 +800,7 @@ async def test_langfuse_trace_id():
litellm.success_callback = ["langfuse"]
litellm_logging_obj = Logging(
model="gpt-3.5-turbo",
model="gpt-5-mini",
messages=[{"role": "user", "content": "hi"}],
stream=False,
call_type="acompletion",
@ -810,7 +810,7 @@ async def test_langfuse_trace_id():
)
litellm.completion(
model="gpt-3.5-turbo",
model="gpt-5-mini",
messages=[{"role": "user", "content": "Hey how's it going?"}],
mock_response="Hey!",
litellm_logging_obj=litellm_logging_obj,

16
tests/logging_callback_tests/test_amazing_s3_logs.py
View file

@ -36,7 +36,7 @@ async def test_basic_s3_logging(sync_mode, streaming):
response_id = None
if sync_mode is True:
response = litellm.completion(
model="gpt-3.5-turbo",
model="gpt-5-mini",
messages=[{"role": "user", "content": "This is a test"}],
mock_response="It's simple to use and easy to get started",
stream=streaming,
@ -50,7 +50,7 @@ async def test_basic_s3_logging(sync_mode, streaming):
time.sleep(2)
else:
response = await litellm.acompletion(
model="gpt-3.5-turbo",
model="gpt-5-mini",
messages=[{"role": "user", "content": "This is a test"}],
mock_response="It's simple to use and easy to get started",
stream=streaming,
@ -102,7 +102,7 @@ async def test_basic_s3_v2_logging(streaming):
litellm.set_verbose = True
response_id = None
response = await litellm.acompletion(
model="gpt-4o-mini",
model="gpt-5-mini",
messages=[{"role": "user", "content": "This is a test"}],
mock_response="It's simple to use and easy to get started",
stream=streaming,
@ -149,7 +149,7 @@ async def test_basic_s3_v2_logging_failure():
# Mock the upload process but still make the httpx call
url = f"https://test-bucket.s3.us-west-2.amazonaws.com/{batch_logging_element.s3_object_key}"
headers = {"Content-Type": "application/json"}
data = '{"model": "gpt-4o-mini"}'
data = '{"model": "gpt-5-mini"}'
# Make the actual httpx call we want to test
await s3_v2_logger.async_httpx_client.put(url=url, headers=headers, data=data)
@ -169,7 +169,7 @@ async def test_basic_s3_v2_logging_failure():
# Trigger a failure by using invalid API key
try:
response = await litellm.acompletion(
model="gpt-4o-mini",
model="gpt-5-mini",
api_key="invalid-api-key",
messages=[{"role": "user", "content": "This is a test"}],
)
@ -203,7 +203,7 @@ async def test_basic_s3_v2_logging_failure():
# Verify JSON data was included
data = call_args[1]["data"]
assert data is not None
assert '"model": "gpt-4o-mini"' in data
assert '"model": "gpt-5-mini"' in data
print("✓ S3 request data contains expected log payload")
@ -256,7 +256,7 @@ def test_s3_logging():
async def _test():
return await litellm.acompletion(
model="gpt-3.5-turbo",
model="gpt-5-mini",
messages=[{"role": "user", "content": f"This is a test {curr_time}"}],
max_tokens=10,
temperature=0.7,
@ -269,7 +269,7 @@ def test_s3_logging():
async def _test():
return await litellm.acompletion(
model="gpt-3.5-turbo",
model="gpt-5-mini",
messages=[{"role": "user", "content": f"This is a test {curr_time}"}],
max_tokens=10,
temperature=0.7,

12
tests/logging_callback_tests/test_assemble_streaming_responses.py
View file

@ -65,7 +65,7 @@ def test_assemble_complete_response_from_streaming_chunks_1(is_async):
)
],
"created": 1721353246,
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"object": "chat.completion.chunk",
"system_fingerprint": None,
"usage": None,
@ -105,7 +105,7 @@ def test_assemble_complete_response_from_streaming_chunks_1(is_async):
)
],
"created": 1721353246,
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"object": "chat.completion.chunk",
"system_fingerprint": None,
"usage": None,
@ -166,7 +166,7 @@ def test_assemble_complete_response_from_streaming_chunks_2(is_async):
)
],
"created": 1721353246,
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"object": "chat.completion.chunk",
"system_fingerprint": None,
"usage": None,
@ -208,7 +208,7 @@ def test_assemble_complete_response_from_streaming_chunks_2(is_async):
)
],
"created": 1721353246,
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"object": "chat.completion.chunk",
"system_fingerprint": None,
"usage": None,
@ -263,7 +263,7 @@ def test_assemble_complete_response_from_streaming_chunks_3(is_async):
)
],
"created": 1721353246,
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"object": "chat.completion.chunk",
"system_fingerprint": None,
"usage": None,
@ -340,7 +340,7 @@ def test_assemble_complete_response_from_streaming_chunks_4(is_async):
)
],
"created": 1721353246,
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"object": "chat.completion.chunk",
"system_fingerprint": None,
"usage": None,

20
tests/logging_callback_tests/test_bedrock_knowledgebase_hook.py
View file

@ -445,7 +445,7 @@ async def test_openai_with_knowledge_base_mock_openai(setup_vector_store_registr
mock_response.id = "chatcmpl-123"
mock_response.object = "chat.completion"
mock_response.created = 1234567890
mock_response.model = "gpt-4"
mock_response.model = "gpt-5.5"
# Store the request for verification
captured_request.update(kwargs)
@ -459,7 +459,7 @@ async def test_openai_with_knowledge_base_mock_openai(setup_vector_store_registr
try:
await litellm.acompletion(
model="gpt-4",
model="gpt-5.5",
messages=[{"role": "user", "content": "what is litellm?"}],
vector_store_ids=["T37J8R4WTM"],
client=client,
@ -521,7 +521,7 @@ async def test_openai_with_vector_store_ids_in_tool_call_mock_openai(
mock_response.id = "chatcmpl-123"
mock_response.object = "chat.completion"
mock_response.created = 1234567890
mock_response.model = "gpt-4"
mock_response.model = "gpt-5.5"
# Store the request for verification
captured_request.update(kwargs)
@ -535,7 +535,7 @@ async def test_openai_with_vector_store_ids_in_tool_call_mock_openai(
try:
await litellm.acompletion(
model="gpt-4",
model="gpt-5.5",
messages=[{"role": "user", "content": "what is litellm?"}],
tools=[{"type": "file_search", "vector_store_ids": ["T37J8R4WTM"]}],
client=client,
@ -594,7 +594,7 @@ async def test_openai_with_mixed_tool_call_mock_openai(setup_vector_store_regist
mock_response.id = "chatcmpl-123"
mock_response.object = "chat.completion"
mock_response.created = 1234567890
mock_response.model = "gpt-4"
mock_response.model = "gpt-5.5"
# Store the request for verification
captured_request.update(kwargs)
@ -608,7 +608,7 @@ async def test_openai_with_mixed_tool_call_mock_openai(setup_vector_store_regist
try:
await litellm.acompletion(
model="gpt-4",
model="gpt-5.5",
messages=[{"role": "user", "content": "what is litellm?"}],
tools=[
{"type": "file_search", "vector_store_ids": ["T37J8R4WTM"]},
@ -642,7 +642,7 @@ async def test_openai_with_mixed_tool_call_mock_openai(setup_vector_store_regist
# test_custom_logger = MockCustomLogger()
# litellm.set_verbose = True
# await litellm.acompletion(
# model="gpt-4",
# model="gpt-5.5",
# messages=[{"role": "user", "content": "what is litellm?"}],
# vector_store_ids = [
# "T37J8R4WTM"
@ -834,7 +834,7 @@ async def test_provider_specific_fields_in_proxy_http_response(
# Initialize proxy
await initialize(
model="gpt-3.5-turbo",
model="gpt-5-mini",
alias=None,
api_base=None,
debug=False,
@ -857,7 +857,7 @@ async def test_provider_specific_fields_in_proxy_http_response(
# Create mock response with provider_specific_fields
mock_response = litellm.ModelResponse(
id="test-123",
model="gpt-3.5-turbo",
model="gpt-5-mini",
created=1234567890,
object="chat.completion",
)
@ -897,7 +897,7 @@ async def test_provider_specific_fields_in_proxy_http_response(
response = client.post(
"/v1/chat/completions",
json={
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"messages": [{"role": "user", "content": "What is litellm?"}],
},
)

20
tests/logging_callback_tests/test_custom_callback_router.py
View file

@ -441,7 +441,7 @@ async def test_async_chat_azure():
# failure
model_list = [
{
"model_name": "gpt-3.5-turbo", # openai model name
"model_name": "gpt-5-mini", # openai model name
"litellm_params": { # params for litellm completion/embedding call
"model": "azure/gpt-4o-new-test",
"api_key": "my-bad-key",
@ -458,7 +458,7 @@ async def test_async_chat_azure():
router3 = Router(model_list=model_list, num_retries=0) # type: ignore
try:
response = await router3.acompletion(
model="gpt-3.5-turbo",
model="gpt-5-mini",
messages=[{"role": "user", "content": "Hi 👋 - i'm openai"}],
)
print(f"response in router3 acompletion: {response}")
@ -547,7 +547,7 @@ async def test_async_chat_azure_with_fallbacks():
# with fallbacks
model_list = [
{
"model_name": "gpt-3.5-turbo", # openai model name
"model_name": "gpt-5-mini", # openai model name
"litellm_params": { # params for litellm completion/embedding call
"model": "azure/gpt-4.1-mini",
"api_key": "my-bad-key",
@ -568,13 +568,13 @@ async def test_async_chat_azure_with_fallbacks():
]
router = Router(
model_list=model_list,
fallbacks=[{"gpt-3.5-turbo": ["gpt-3.5-turbo-16k"]}],
fallbacks=[{"gpt-5-mini": ["gpt-3.5-turbo-16k"]}],
retry_policy=litellm.router.RetryPolicy(
AuthenticationErrorRetries=0,
),
) # type: ignore
response = await router.acompletion(
model="gpt-3.5-turbo",
model="gpt-5-mini",
messages=[{"role": "user", "content": "Hi 👋 - i'm openai"}],
)
await asyncio.sleep(2)
@ -731,9 +731,9 @@ async def test_async_embedding_azure_caching():
router = Router(
model_list=[
{
"model_name": "text-embedding-ada-002",
"model_name": "text-embedding-3-small",
"litellm_params": {
"model": "openai/text-embedding-ada-002",
"model": "openai/text-embedding-3-small",
},
}
]
@ -741,13 +741,13 @@ async def test_async_embedding_azure_caching():
litellm.callbacks = [customHandler_caching]
unique_time = time.time()
response1 = await router.aembedding(
model="text-embedding-ada-002",
model="text-embedding-3-small",
input=[f"good morning from litellm1 {unique_time}"],
caching=True,
)
await asyncio.sleep(1) # set cache is async for aembedding()
response2 = await router.aembedding(
model="text-embedding-ada-002",
model="text-embedding-3-small",
input=[f"good morning from litellm1 {unique_time}"],
caching=True,
)
@ -776,7 +776,7 @@ async def test_rate_limit_error_callback():
{
"model_name": "my-test-gpt",
"litellm_params": {
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"mock_response": "litellm.RateLimitError",
},
}

12
tests/logging_callback_tests/test_datadog.py
View file

@ -54,9 +54,9 @@ def create_standard_logging_payload() -> StandardLoggingPayload:
endTime=1234567891.0,
completionStartTime=1234567890.5,
model_map_information=StandardLoggingModelInformation(
model_map_key="gpt-3.5-turbo", model_map_value=None
model_map_key="gpt-4.1-mini", model_map_value=None
),
model="gpt-3.5-turbo",
model="gpt-4.1-mini",
model_id="model-123",
model_group="openai-gpt",
api_base="https://api.openai.com",
@ -195,7 +195,7 @@ async def test_datadog_logging_http_request():
# Make the completion call
for _ in range(5):
response = await litellm.acompletion(
model="gpt-3.5-turbo",
model="gpt-4.1-mini",
messages=[{"role": "user", "content": "what llm are u"}],
max_tokens=10,
temperature=0.2,
@ -279,7 +279,7 @@ async def test_datadog_logging_http_request():
# Check specific fields
assert message["call_type"] == "acompletion"
assert message["model"] == "gpt-3.5-turbo"
assert message["model"] == "gpt-4.1-mini"
assert isinstance(message["model_parameters"], dict)
assert "temperature" in message["model_parameters"]
assert "max_tokens" in message["model_parameters"]
@ -411,7 +411,7 @@ async def test_datadog_log_redis_failures():
# Make the completion call
for _ in range(3):
response = await litellm.acompletion(
model="gpt-3.5-turbo",
model="gpt-4.1-mini",
messages=[{"role": "user", "content": "what llm are u"}],
max_tokens=10,
temperature=0.2,
@ -469,7 +469,7 @@ async def test_datadog_logging():
litellm.success_callback = ["datadog"]
litellm.set_verbose = True
response = await litellm.acompletion(
model="gpt-3.5-turbo",
model="gpt-4.1-mini",
messages=[{"role": "user", "content": "what llm are u"}],
max_tokens=10,
temperature=0.2,

6
tests/logging_callback_tests/test_datadog_llm_obs.py
View file

@ -48,9 +48,9 @@ def create_standard_logging_payload() -> StandardLoggingPayload:
endTime=1234567891.0,
completionStartTime=1234567890.5,
model_map_information=StandardLoggingModelInformation(
model_map_key="gpt-3.5-turbo", model_map_value=None
model_map_key="gpt-5-mini", model_map_value=None
),
model="gpt-3.5-turbo",
model="gpt-5-mini",
model_id="model-123",
model_group="openai-gpt",
api_base="https://api.openai.com",
@ -93,7 +93,7 @@ async def test_datadog_llm_obs_logging():
for _ in range(2):
response = await litellm.acompletion(
model="gpt-4o",
model="gpt-5.5",
messages=[{"role": "user", "content": "Hello testing dd llm obs!"}],
mock_response="hi",
)

26
tests/logging_callback_tests/test_generic_api_callback.py
View file

@ -59,7 +59,7 @@ async def test_generic_api_callback():
# Make the completion call
response = await litellm.acompletion(
model="gpt-4o",
model="gpt-5.5",
messages=[{"role": "user", "content": "Hello, world!"}],
mock_response="hi",
user="test_user",
@ -109,11 +109,11 @@ async def test_generic_api_callback():
# Basic assertions for standard logging payload
assert payload_item["response_cost"] > 0, "Response cost should be greater than 0"
assert payload_item["model"] == "gpt-4o", "Model should be gpt-4o"
assert payload_item["model"] == "gpt-5.5", "Model should be gpt-5.5"
assert (
payload_item["model_parameters"]["user"] == "test_user"
), "User should be test_user"
assert payload_item["model"] == "gpt-4o", "Model should be gpt-4o"
assert payload_item["model"] == "gpt-5.5", "Model should be gpt-5.5"
assert payload_item["messages"] == [
{"role": "user", "content": "Hello, world!"}
], "Messages should be the same"
@ -147,7 +147,7 @@ async def test_generic_api_callback_multiple_logs():
# Make the completion call
for _ in range(10):
response = await litellm.acompletion(
model="gpt-4o",
model="gpt-5.5",
messages=[{"role": "user", "content": "Hello, world!"}],
mock_response="hi",
user="test_user",
@ -197,11 +197,11 @@ async def test_generic_api_callback_multiple_logs():
assert (
payload_item["response_cost"] > 0
), "Response cost should be greater than 0"
assert payload_item["model"] == "gpt-4o", "Model should be gpt-4o"
assert payload_item["model"] == "gpt-5.5", "Model should be gpt-5.5"
assert (
payload_item["model_parameters"]["user"] == "test_user"
), "User should be test_user"
assert payload_item["model"] == "gpt-4o", "Model should be gpt-4o"
assert payload_item["model"] == "gpt-5.5", "Model should be gpt-5.5"
assert payload_item["messages"] == [
{"role": "user", "content": "Hello, world!"}
], "Messages should be the same"
@ -239,7 +239,7 @@ async def test_generic_api_callback_ndjson_format():
# Make multiple completion calls to generate multiple logs
for i in range(3):
response = await litellm.acompletion(
model="gpt-4o",
model="gpt-5.5",
messages=[{"role": "user", "content": f"Hello, world! {i}"}],
mock_response="hi",
user="test_user",
@ -279,7 +279,7 @@ async def test_generic_api_callback_ndjson_format():
assert (
payload_item["response_cost"] > 0
), "Response cost should be greater than 0"
assert payload_item["model"] == "gpt-4o", "Model should be gpt-4o"
assert payload_item["model"] == "gpt-5.5", "Model should be gpt-5.5"
assert (
payload_item["model_parameters"]["user"] == "test_user"
), "User should be test_user"
@ -314,7 +314,7 @@ async def test_generic_api_callback_single_format():
# Make 3 completion calls
for i in range(3):
response = await litellm.acompletion(
model="gpt-4o",
model="gpt-5.5",
messages=[{"role": "user", "content": f"Hello, world! {i}"}],
mock_response="hi",
user="test_user",
@ -345,7 +345,7 @@ async def test_generic_api_callback_single_format():
assert (
payload_item["response_cost"] > 0
), "Response cost should be greater than 0"
assert payload_item["model"] == "gpt-4o", "Model should be gpt-4o"
assert payload_item["model"] == "gpt-5.5", "Model should be gpt-5.5"
@pytest.mark.asyncio
@ -377,7 +377,7 @@ async def test_generic_api_callback_json_array_format_explicit():
# Make multiple completion calls
for i in range(5):
response = await litellm.acompletion(
model="gpt-4o",
model="gpt-5.5",
messages=[{"role": "user", "content": f"Hello, world! {i}"}],
mock_response="hi",
user="test_user",
@ -404,7 +404,7 @@ async def test_generic_api_callback_json_array_format_explicit():
assert (
payload_item["response_cost"] > 0
), "Response cost should be greater than 0"
assert payload_item["model"] == "gpt-4o", "Model should be gpt-4o"
assert payload_item["model"] == "gpt-5.5", "Model should be gpt-5.5"
@pytest.mark.asyncio
@ -434,7 +434,7 @@ async def test_generic_api_callback_sumologic_uses_ndjson():
# Make completion calls
for i in range(2):
await litellm.acompletion(
model="gpt-4o",
model="gpt-5.5",
messages=[{"role": "user", "content": f"Test {i}"}],
mock_response="response",
user="test_user",

4
tests/logging_callback_tests/test_langfuse_unit_tests.py
View file

@ -40,9 +40,9 @@ def create_standard_logging_payload() -> StandardLoggingPayload:
endTime=1234567891.0,
completionStartTime=1234567890.5,
model_map_information=StandardLoggingModelInformation(
model_map_key="gpt-3.5-turbo", model_map_value=None
model_map_key="gpt-5-mini", model_map_value=None
),
model="gpt-3.5-turbo",
model="gpt-5-mini",
model_id="model-123",
model_group="openai-gpt",
api_base="https://api.openai.com",

10
tests/logging_callback_tests/test_langsmith_unit_test.py
View file

@ -332,7 +332,7 @@ async def test_langsmith_key_based_logging():
litellm.callbacks = [LangsmithLogger()]
response = await litellm.acompletion(
model="gpt-3.5-turbo",
model="gpt-4.1-mini",
messages=[{"role": "user", "content": "Test message"}],
max_tokens=10,
temperature=0.2,
@ -373,7 +373,7 @@ async def test_langsmith_key_based_logging():
"inputs": {
"id": "chatcmpl-82699ee4-7932-4fc0-9585-76abc8caeafa",
"call_type": "acompletion",
"model": "gpt-3.5-turbo",
"model": "gpt-4.1-mini",
"messages": [{"role": "user", "content": "Test message"}],
"model_parameters": {
"temperature": 0.2,
@ -382,7 +382,7 @@ async def test_langsmith_key_based_logging():
},
"outputs": {
"id": "chatcmpl-82699ee4-7932-4fc0-9585-76abc8caeafa",
"model": "gpt-3.5-turbo",
"model": "gpt-4.1-mini",
"choices": [
{
"finish_reason": "stop",
@ -468,7 +468,7 @@ async def test_langsmith_queue_logging():
# Make multiple calls to ensure we don't hit the batch size
for _ in range(5):
response = await litellm.acompletion(
model="gpt-3.5-turbo",
model="gpt-4.1-mini",
messages=[{"role": "user", "content": "Test message"}],
max_tokens=10,
temperature=0.2,
@ -487,7 +487,7 @@ async def test_langsmith_queue_logging():
# Now make calls to exceed the batch size
for _ in range(3):
response = await litellm.acompletion(
model="gpt-3.5-turbo",
model="gpt-4.1-mini",
messages=[{"role": "user", "content": "Test message"}],
max_tokens=10,
temperature=0.2,

16
tests/logging_callback_tests/test_logging_redaction_e2e_test.py
View file

@ -39,7 +39,7 @@ async def test_global_redaction_on():
test_custom_logger = TestCustomLogger()
litellm.callbacks = [test_custom_logger]
response = await litellm.acompletion(
model="gpt-3.5-turbo",
model="gpt-5-mini",
messages=[{"role": "user", "content": "hi"}],
mock_response="hello",
)
@ -69,7 +69,7 @@ async def test_global_redaction_ignores_dynamic_param(turn_off_message_logging):
test_custom_logger = TestCustomLogger()
litellm.callbacks = [test_custom_logger]
response = await litellm.acompletion(
model="gpt-3.5-turbo",
model="gpt-5-mini",
messages=[{"role": "user", "content": "hi"}],
turn_off_message_logging=turn_off_message_logging,
mock_response="hello",
@ -101,7 +101,7 @@ async def test_global_redaction_off_ignores_dynamic_param(turn_off_message_loggi
test_custom_logger = TestCustomLogger()
litellm.callbacks = [test_custom_logger]
response = await litellm.acompletion(
model="gpt-3.5-turbo",
model="gpt-5-mini",
messages=[{"role": "user", "content": "hi"}],
turn_off_message_logging=turn_off_message_logging,
mock_response="hello",
@ -129,7 +129,7 @@ async def test_redaction_responses_api():
litellm.callbacks = [test_custom_logger]
response = await litellm.aresponses(
model="gpt-3.5-turbo",
model="gpt-5-mini",
input="hi",
mock_response="This is a test response",
)
@ -198,7 +198,7 @@ async def test_redaction_responses_api_stream():
new=mock_post,
):
response = await litellm.aresponses(
model="gpt-3.5-turbo",
model="gpt-5-mini",
input="hi",
stream=True,
)
@ -411,7 +411,7 @@ async def test_redaction_with_streaming_response():
# This simulates the scenario where a streaming response returns a coroutine
# that would normally cause the pickle error
response = await litellm.acompletion(
model="gpt-3.5-turbo",
model="gpt-5-mini",
messages=[{"role": "user", "content": "hi"}],
stream=True,
mock_response="hello",
@ -450,7 +450,7 @@ async def test_disable_redaction_header_responses_api():
# Pass the header via litellm_metadata (as the proxy does for Responses API)
response = await litellm.aresponses(
model="gpt-3.5-turbo",
model="gpt-5-mini",
input="hi",
mock_response="This is a test response",
litellm_metadata={"headers": {"litellm-disable-message-redaction": "true"}},
@ -487,7 +487,7 @@ async def test_redaction_with_metadata_completion_api():
# to determine which field to check. No headers means redaction should happen
# based on the global setting (litellm.turn_off_message_logging = True)
response = await litellm.acompletion(
model="gpt-3.5-turbo",
model="gpt-5-mini",
messages=[{"role": "user", "content": "hi"}],
mock_response="hello",
metadata={},

2
tests/logging_callback_tests/test_opentelemetry_unit_tests.py
View file

@ -53,7 +53,7 @@ class TestOpentelemetryUnitTests(BaseLoggingCallbackTest):
litellm.callbacks = ["otel"]
await litellm.acompletion(
model="gpt-3.5-turbo",
model="gpt-5-mini",
messages=[{"role": "user", "content": "Hello, world!"}],
mock_response="Hey!",
metadata={"litellm_parent_otel_span": parent_otel_span},

8
tests/logging_callback_tests/test_otel_logging.py
View file

@ -48,7 +48,7 @@ async def test_async_otel_callback(streaming):
litellm.callbacks = [OpenTelemetry(config=OpenTelemetryConfig(exporter=exporter))]
response = await litellm.acompletion(
model="gpt-3.5-turbo",
model="gpt-4.1-mini",
messages=[{"role": "user", "content": "hi"}],
temperature=0.1,
user="OTEL_USER",
@ -76,7 +76,7 @@ async def test_async_otel_callback(streaming):
if span.name == "litellm_request":
validate_litellm_request(span)
# Additional specific checks
assert span._attributes["gen_ai.request.model"] == "gpt-3.5-turbo"
assert span._attributes["gen_ai.request.model"] == "gpt-4.1-mini"
assert span._attributes["gen_ai.system"] == "openai"
assert span._attributes["gen_ai.request.temperature"] == 0.1
assert span._attributes["llm.is_streaming"] == str(streaming)
@ -185,7 +185,7 @@ async def test_awesome_otel_with_message_logging_off(streaming, global_redact):
litellm.failure_callback = []
response = await litellm.acompletion(
model="gpt-3.5-turbo",
model="gpt-4.1-mini",
messages=[{"role": "user", "content": "hi"}],
mock_response="hi",
stream=streaming,
@ -293,7 +293,7 @@ async def test_arize_phoenix_creates_nested_spans_on_dedicated_provider():
# Simulate a proxy request by injecting proxy_server_request as a top-level kwarg.
# This triggers ArizePhoenixLogger._get_phoenix_context to create its own parent span.
await litellm.acompletion(
model="gpt-3.5-turbo",
model="gpt-4.1-mini",
messages=[{"role": "user", "content": "ping"}],
mock_response="pong",
proxy_server_request={

10
tests/logging_callback_tests/test_pagerduty_alerting.py
View file

@ -27,7 +27,7 @@ async def test_pagerduty_alerting():
try:
await litellm.acompletion(
model="gpt-3.5-turbo",
model="gpt-5-mini",
messages=[{"role": "user", "content": "hi"}],
mock_response="litellm.RateLimitError",
)
@ -48,7 +48,7 @@ async def test_pagerduty_alerting_high_failure_rate():
try:
await litellm.acompletion(
model="gpt-3.5-turbo",
model="gpt-5-mini",
messages=[{"role": "user", "content": "hi"}],
mock_response="litellm.RateLimitError",
)
@ -61,7 +61,7 @@ async def test_pagerduty_alerting_high_failure_rate():
for _ in range(3):
try:
await litellm.acompletion(
model="gpt-3.5-turbo",
model="gpt-5-mini",
messages=[{"role": "user", "content": "hi"}],
mock_response="litellm.RateLimitError",
)
@ -88,12 +88,12 @@ async def test_pagerduty_hanging_request_alerting():
user_id="test-user",
end_user_id="test-end-user",
),
data={"model": "gpt-4o", "messages": [{"role": "user", "content": "hi"}]},
data={"model": "gpt-5.5", "messages": [{"role": "user", "content": "hi"}]},
call_type="completion",
)
await litellm.acompletion(
model="gpt-4o",
model="gpt-5.5",
messages=[{"role": "user", "content": "hi"}],
)

8
tests/logging_callback_tests/test_posthog.py
View file

@ -33,7 +33,7 @@ def create_standard_logging_payload() -> StandardLoggingPayload:
"endTime": 1234567891.0,
"completionStartTime": 1234567890.5,
"response_time": 1.0,
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"model_id": "model-123",
"api_base": "https://api.openai.com",
"cache_hit": False,
@ -57,7 +57,7 @@ async def test_create_posthog_event_payload():
event_payload = posthog_logger.create_posthog_event_payload(kwargs)
assert event_payload["event"] == "$ai_generation"
assert event_payload["properties"]["$ai_model"] == "gpt-3.5-turbo"
assert event_payload["properties"]["$ai_model"] == "gpt-5-mini"
assert event_payload["properties"]["$ai_input_tokens"] == 20
assert event_payload["properties"]["$ai_output_tokens"] == 10
@ -251,7 +251,7 @@ async def test_custom_metadata_with_no_metadata():
# Should not error and should have standard properties
assert event_payload["event"] == "$ai_generation"
assert event_payload["properties"]["$ai_model"] == "gpt-3.5-turbo"
assert event_payload["properties"]["$ai_model"] == "gpt-5-mini"
# Test with empty metadata
kwargs = {
@ -262,7 +262,7 @@ async def test_custom_metadata_with_no_metadata():
# Should not error and should have standard properties
assert event_payload["event"] == "$ai_generation"
assert event_payload["properties"]["$ai_model"] == "gpt-3.5-turbo"
assert event_payload["properties"]["$ai_model"] == "gpt-5-mini"
@pytest.mark.asyncio

20
tests/logging_callback_tests/test_spend_logs.py
View file

@ -91,7 +91,7 @@ def test_spend_logs_payload(model_id: Optional[str]):
"content-length": "163",
},
"endpoint": "http://localhost:4000/chat/completions",
"model_group": "gpt-3.5-turbo",
"model_group": "gpt-5-mini",
"deployment": "azure/gpt-4.1-mini",
"model_info": {
"id": "4bad40a1eb6bebd1682800f16f44b9f06c52a6703444c99c7f9f32e9de3693b4",
@ -129,7 +129,7 @@ def test_spend_logs_payload(model_id: Optional[str]):
},
{"role": "user", "content": "bom dia"},
],
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"max_tokens": 10,
},
},
@ -332,7 +332,7 @@ def test_spend_logs_payload_with_prompts_enabled(monkeypatch):
input_args: dict = {
"kwargs": {
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"messages": [{"role": "user", "content": "Hello!"}],
"litellm_params": {
"metadata": {
@ -349,7 +349,7 @@ def test_spend_logs_payload_with_prompts_enabled(monkeypatch):
message=litellm.Message(content="Hi there!", role="assistant"),
)
],
model="gpt-3.5-turbo",
model="gpt-5-mini",
usage=litellm.Usage(completion_tokens=2, prompt_tokens=1, total_tokens=3),
),
"start_time": datetime.datetime.now(),
@ -372,7 +372,7 @@ def test_spend_logs_payload_with_prompts_enabled(monkeypatch):
litellm_params = {
"proxy_server_request": {
"body": {
"model": "gpt-4",
"model": "gpt-5.5",
"messages": [{"role": "user", "content": "Hello!"}],
}
}
@ -389,7 +389,7 @@ def test_spend_logs_payload_with_prompts_enabled(monkeypatch):
{"role": "assistant", "content": "Hi there!"}
)
proxy_server_request = json.loads(payload["proxy_server_request"] or "{}")
assert proxy_server_request["model"] == "gpt-4"
assert proxy_server_request["model"] == "gpt-5.5"
assert proxy_server_request["messages"] == [{"role": "user", "content": "Hello!"}]
# Clean up - reset general_settings
@ -420,7 +420,7 @@ def test_large_request_no_truncation_threshold():
request_body = {
"messages": [{"role": "user", "content": large_content}],
"model": "gpt-4",
"model": "gpt-5.5",
}
sanitized = _sanitize_request_body_for_spend_logs_payload(request_body)
@ -454,7 +454,7 @@ def test_small_request_no_truncation():
request_body = {
"messages": [{"role": "user", "content": small_content}],
"model": "gpt-4",
"model": "gpt-5.5",
}
sanitized = _sanitize_request_body_for_spend_logs_payload(request_body)
@ -497,7 +497,7 @@ def test_configurable_string_length_env_var(monkeypatch):
request_body = {
"messages": [{"role": "user", "content": large_content}],
"model": "gpt-4",
"model": "gpt-5.5",
}
sanitized = _sanitize_request_body_for_spend_logs_payload(request_body)
@ -531,7 +531,7 @@ def test_truncation_preserves_beginning_and_end():
request_body = {
"messages": [{"role": "user", "content": large_content}],
"model": "gpt-4",
"model": "gpt-5.5",
}
sanitized = _sanitize_request_body_for_spend_logs_payload(request_body)

8
tests/logging_callback_tests/test_sqs_logger.py
View file

@ -34,7 +34,7 @@ async def test_async_sqs_logger_flush():
litellm.callbacks = [sqs_logger]
await litellm.acompletion(
model="gpt-4o",
model="gpt-5.5",
messages=[{"role": "user", "content": "hello"}],
mock_response="hi",
)
@ -74,7 +74,7 @@ async def test_async_sqs_logger_flush():
assert "model" in payload_data
assert "messages" in payload_data
assert "response" in payload_data
assert payload_data["model"] == "gpt-4o"
assert payload_data["model"] == "gpt-5.5"
assert len(payload_data["messages"]) == 1
assert payload_data["messages"][0]["role"] == "user"
assert payload_data["messages"][0]["content"] == "hello"
@ -99,7 +99,7 @@ async def test_async_sqs_logger_error_flush():
litellm.callbacks = [sqs_logger]
await litellm.acompletion(
model="gpt-4o",
model="gpt-5.5",
messages=[{"role": "user", "content": "hello"}],
mock_response="Error occurred",
)
@ -139,7 +139,7 @@ async def test_async_sqs_logger_error_flush():
assert "model" in payload_data
assert "messages" in payload_data
assert "response" in payload_data
assert payload_data["model"] == "gpt-4o"
assert payload_data["model"] == "gpt-5.5"
assert len(payload_data["messages"]) == 1
assert payload_data["messages"][0]["role"] == "user"
assert payload_data["messages"][0]["content"] == "hello"

26
tests/logging_callback_tests/test_standard_logging_payload.py
View file

@ -317,16 +317,16 @@ def test_get_model_cost_information():
# Test with valid model
result = StandardLoggingPayloadSetup.get_model_cost_information(
base_model="gpt-3.5-turbo",
base_model="gpt-5-mini",
custom_pricing=False,
custom_llm_provider="openai",
init_response_obj={},
)
litellm_info_gpt_3_5_turbo_model_map_value = litellm.get_model_info(
model="gpt-3.5-turbo", custom_llm_provider="openai"
model="gpt-5-mini", custom_llm_provider="openai"
)
print("result", result)
assert result["model_map_key"] == "gpt-3.5-turbo"
assert result["model_map_key"] == "gpt-5-mini"
assert result["model_map_value"] is not None
assert result["model_map_value"] == litellm_info_gpt_3_5_turbo_model_map_value
# assert all fields in StandardLoggingModelInformation are present
@ -515,7 +515,7 @@ def test_get_error_information():
litellm_exception = litellm.exceptions.RateLimitError(
message="Test error",
llm_provider="openai",
model="gpt-3.5-turbo",
model="gpt-5-mini",
response=None,
litellm_debug_info=None,
max_retries=None,
@ -603,7 +603,7 @@ def test_cost_breakdown_in_standard_logging_payload():
# Create a mock logging object with cost breakdown
logging_obj = Logging(
model="gpt-4o",
model="gpt-5.5",
messages=[{"role": "user", "content": "Hello"}],
stream=False,
call_type="completion",
@ -624,7 +624,7 @@ def test_cost_breakdown_in_standard_logging_payload():
mock_response = {
"id": "chatcmpl-123",
"object": "chat.completion",
"model": "gpt-4o",
"model": "gpt-5.5",
"usage": {
"prompt_tokens": 10,
"completion_tokens": 20,
@ -644,7 +644,7 @@ def test_cost_breakdown_in_standard_logging_payload():
# Create kwargs
kwargs = {
"model": "gpt-4o",
"model": "gpt-5.5",
"messages": [{"role": "user", "content": "Hello"}],
"response_cost": 0.0035,
"custom_llm_provider": "openai",
@ -687,7 +687,7 @@ def test_cost_breakdown_missing_in_standard_logging_payload():
# Create a mock logging object without cost breakdown
logging_obj = Logging(
model="gpt-4o",
model="gpt-5.5",
messages=[{"role": "user", "content": "Hello"}],
stream=False,
call_type="embedding", # Non-completion call type
@ -702,12 +702,12 @@ def test_cost_breakdown_missing_in_standard_logging_payload():
mock_response = {
"object": "list",
"data": [{"embedding": [0.1, 0.2, 0.3]}],
"model": "text-embedding-ada-002",
"model": "text-embedding-3-small",
"usage": {"prompt_tokens": 10, "total_tokens": 10},
}
kwargs = {
"model": "text-embedding-ada-002",
"model": "text-embedding-3-small",
"input": ["Hello"],
"response_cost": 0.0001,
"custom_llm_provider": "openai",
@ -756,7 +756,7 @@ def test_usage_dict_roundtrip_in_payload(use_combined_usage_object):
from datetime import datetime
logging_obj = Logging(
model="gpt-4o",
model="gpt-5.5",
messages=[{"role": "user", "content": "Hi"}],
stream=False,
call_type="completion",
@ -768,7 +768,7 @@ def test_usage_dict_roundtrip_in_payload(use_combined_usage_object):
mock_response = {
"id": "chatcmpl-usage-test",
"object": "chat.completion",
"model": "gpt-4o",
"model": "gpt-5.5",
"usage": {
"prompt_tokens": 42,
"completion_tokens": 58,
@ -784,7 +784,7 @@ def test_usage_dict_roundtrip_in_payload(use_combined_usage_object):
}
kwargs = {
"model": "gpt-4o",
"model": "gpt-5.5",
"messages": [{"role": "user", "content": "Hi"}],
"response_cost": 0.01,
"custom_llm_provider": "openai",

2
tests/logging_callback_tests/test_standard_logging_payload_excluded_fields.py
View file

@ -49,7 +49,7 @@ def create_sample_standard_logging_payload() -> Dict:
"completionStartTime": 1234567890.5,
"response_time": 1.0,
"model_map_information": {},
"model": "gpt-4",
"model": "gpt-5.5",
"model_id": "model-123",
"model_group": None,
"api_base": "https://api.openai.com/v1",

6
tests/logging_callback_tests/test_token_counting.py
View file

@ -55,7 +55,7 @@ async def test_stream_token_counting_gpt_4o():
litellm.logging_callback_manager.add_litellm_callback(custom_logger)
response = await litellm.acompletion(
model="gpt-4o",
model="gpt-5.5",
messages=[{"role": "user", "content": "Hello, how are you?" * 100}],
stream=True,
stream_options={"include_usage": True},
@ -95,7 +95,7 @@ async def test_stream_token_counting_without_include_usage():
litellm.logging_callback_manager.add_litellm_callback(custom_logger)
response = await litellm.acompletion(
model="gpt-4o",
model="gpt-5.5",
messages=[{"role": "user", "content": "Hello, how are you?" * 100}],
stream=True,
)
@ -133,7 +133,7 @@ async def test_stream_token_counting_with_redaction():
litellm.logging_callback_manager.add_litellm_callback(custom_logger)
response = await litellm.acompletion(
model="gpt-4o",
model="gpt-5.5",
messages=[{"role": "user", "content": "Hello, how are you?" * 100}],
stream=True,
)

2
tests/logging_callback_tests/test_unit_test_litellm_logging.py
View file

@ -27,7 +27,7 @@ service_logger = ServiceLogging()
def setup_logging():
return Logging(
model="gpt-4o",
model="gpt-5.5",
messages=[{"role": "user", "content": "Hello, world!"}],
stream=False,
call_type="completion",

10
tests/logging_callback_tests/test_unit_tests_init_callbacks.py
View file

@ -164,7 +164,7 @@ async def use_callback_in_llm_call(
for _ in range(5):
await litellm.acompletion(
model="gpt-3.5-turbo",
model="gpt-5-mini",
messages=[{"role": "user", "content": "hi"}],
temperature=0.1,
mock_response="hello",
@ -217,7 +217,7 @@ def test_dynamic_logging_global_callback():
cl = CustomLogger()
litellm_logging = LiteLLMLoggingObj(
model="claude-3-opus-20240229",
model="claude-opus-4-7",
messages=[{"role": "user", "content": "hi"}],
stream=False,
call_type="completion",
@ -240,7 +240,7 @@ def test_dynamic_logging_global_callback():
result=ModelResponse(
id="chatcmpl-5418737b-ab14-420b-b9c5-b278b6681b70",
created=1732306261,
model="claude-3-opus-20240229",
model="claude-opus-4-7",
object="chat.completion",
system_fingerprint=None,
choices=[
@ -277,7 +277,7 @@ def test_get_combined_callback_list():
from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj
_logging = LiteLLMLoggingObj(
model="claude-3-opus-20240229",
model="claude-opus-4-7",
messages=[{"role": "user", "content": "hi"}],
stream=False,
call_type="completion",
@ -298,7 +298,7 @@ def test_get_combined_callback_list_returns_copy_when_dynamic_is_none():
from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj
_logging = LiteLLMLoggingObj(
model="claude-3-opus-20240229",
model="claude-opus-4-7",
messages=[{"role": "user", "content": "hi"}],
stream=False,
call_type="completion",

10
tests/openai_endpoints_tests/test_e2e_openai_responses_api.py
View file

@ -74,7 +74,7 @@ def validate_stream_chunk(chunk):
def test_basic_response():
client = get_test_client()
response = client.responses.create(
model="gpt-4o", input="just respond with the word 'ping'"
model="gpt-5.5", input="just respond with the word 'ping'"
)
print("basic response=", response)
@ -94,7 +94,7 @@ def test_basic_response():
def test_streaming_response():
client = get_test_client()
stream = client.responses.create(
model="gpt-4o", input="just respond with the word 'ping'", stream=True
model="gpt-5.5", input="just respond with the word 'ping'", stream=True
)
collected_chunks = []
@ -117,7 +117,7 @@ def test_bad_request_bad_param_error():
with pytest.raises(BadRequestError):
# Trigger error with invalid model name
client.responses.create(
model="gpt-4o", input="This should fail", temperature=2000
model="gpt-5.5", input="This should fail", temperature=2000
)
@ -137,7 +137,7 @@ def test_cancel_response():
from litellm.types.llms.openai import ResponsesAPIResponse
response = client.responses.create(
model="gpt-4o", input="just respond with the word 'ping'", background=True
model="gpt-5.5", input="just respond with the word 'ping'", background=True
)
print("basic response=", response)
@ -160,7 +160,7 @@ def test_cancel_streaming_response():
from litellm.types.llms.openai import ResponsesAPIResponse
stream = client.responses.create(
model="gpt-4o",
model="gpt-5.5",
input="just respond with the word 'ping'",
stream=True,
background=True,

4
tests/openai_endpoints_tests/test_openai_batches_endpoint.py
View file

@ -233,8 +233,8 @@ async def test_list_batches_with_target_model_names():
"""
# Test data
target_model_names = "gpt-4,gpt-3.5-turbo"
expected_model = "gpt-4" # Should use the first model from the comma-separated list
target_model_names = "gpt-5.5,gpt-5-mini"
expected_model = "gpt-5.5" # Should use the first model from the comma-separated list
# Mock response for list_batches
mock_batch_response = {

2
tests/openai_endpoints_tests/test_openai_fine_tuning.py
View file

@ -30,7 +30,7 @@ async def test_openai_fine_tuning():
# create fine tuning job
ft_job = await client.fine_tuning.jobs.create(
model="gpt-4o-mini-2024-07-18",
model="gpt-4.1-mini-2025-04-14",
training_file=response.id,
extra_headers={"custom-llm-provider": "openai"},
)

4
tests/openai_endpoints_tests/test_responses_websocket_proxy_e2e.py
View file

@ -6,7 +6,7 @@ and validates the streamed response events.
Requires:
- Proxy running: python -m litellm.proxy.proxy_cli --config <config> --port 4000
- Model configured in proxy (e.g. gpt-4o-mini)
- Model configured in proxy (e.g. gpt-5-mini)
See: https://developers.openai.com/api/docs/guides/websocket-mode/
"""
@ -21,7 +21,7 @@ import pytest
# ── Configuration ─────────────────────────────────────────────────────────────
PROXY_BASE_URL = os.environ.get("LITELLM_PROXY_BASE_URL", "ws://0.0.0.0:4000")
PROXY_MASTER_KEY = os.environ.get("LITELLM_PROXY_KEY", "sk-1234")
PROXY_MODEL = os.environ.get("LITELLM_PROXY_RESPONSES_MODEL", "gpt-4o-mini")
PROXY_MODEL = os.environ.get("LITELLM_PROXY_RESPONSES_MODEL", "gpt-5-mini")
# ──────────────────────────────────────────────────────────────────────────────

34
tests/otel_tests/test_e2e_model_access.py
View file

@ -59,12 +59,12 @@ async def mock_chat_completion(session, key: str, model: str):
"key_models, test_model, expect_success",
[
(["openai/*"], "anthropic/claude-2", False), # Non-matching model
(["gpt-4"], "gpt-4", True), # Exact model match
(["gpt-5.5"], "gpt-5.5", True), # Exact model match
(["bedrock/*"], "bedrock/anthropic.claude-3", True), # Bedrock wildcard
(["bedrock/anthropic.*"], "bedrock/anthropic.claude-3", True), # Pattern match
(["bedrock/anthropic.*"], "bedrock/amazon.titan", False), # Pattern non-match
(None, "gpt-4", True), # No model restrictions
([], "gpt-4", True), # Empty model list
(None, "gpt-5.5", True), # No model restrictions
([], "gpt-5.5", True), # Empty model list
],
)
@pytest.mark.asyncio
@ -119,7 +119,7 @@ async def test_model_access_update():
response = await client.post(
"/key/generate",
json={
"models": ["openai/gpt-4"],
"models": ["openai/gpt-5.5"],
"metadata": dict(_ALLOW_CLIENT_MOCK_METADATA),
},
headers=headers,
@ -130,13 +130,13 @@ async def test_model_access_update():
# Test initial access
async with aiohttp.ClientSession() as session:
# Should work with gpt-4
await mock_chat_completion(session=session, key=key, model="openai/gpt-4")
# Should work with gpt-5.5
await mock_chat_completion(session=session, key=key, model="openai/gpt-5.5")
# Should fail with gpt-3.5-turbo
# Should fail with gpt-5-mini
with pytest.raises(Exception) as exc_info:
await mock_chat_completion(
session=session, key=key, model="openai/gpt-3.5-turbo"
session=session, key=key, model="openai/gpt-5-mini"
)
_validate_model_access_exception(
exc_info.value, expected_type="key_model_access_denied"
@ -151,9 +151,9 @@ async def test_model_access_update():
# Test updated access
async with aiohttp.ClientSession() as session:
# Both models should now work
await mock_chat_completion(session=session, key=key, model="openai/gpt-4")
await mock_chat_completion(session=session, key=key, model="openai/gpt-5.5")
await mock_chat_completion(
session=session, key=key, model="openai/gpt-3.5-turbo"
session=session, key=key, model="openai/gpt-5-mini"
)
# Non-OpenAI model should still fail
@ -226,7 +226,7 @@ async def test_team_model_access_update():
response = await client.post(
"/team/new",
json={
"models": ["openai/gpt-4"],
"models": ["openai/gpt-5.5"],
"name": "test-team",
"metadata": dict(_ALLOW_CLIENT_MOCK_METADATA),
},
@ -250,13 +250,13 @@ async def test_team_model_access_update():
# Test initial access
async with aiohttp.ClientSession() as session:
# Should work with gpt-4
await mock_chat_completion(session=session, key=key, model="openai/gpt-4")
# Should work with gpt-5.5
await mock_chat_completion(session=session, key=key, model="openai/gpt-5.5")
# Should fail with gpt-3.5-turbo
# Should fail with gpt-5-mini
with pytest.raises(Exception) as exc_info:
await mock_chat_completion(
session=session, key=key, model="openai/gpt-3.5-turbo"
session=session, key=key, model="openai/gpt-5-mini"
)
_validate_model_access_exception(
exc_info.value, expected_type="team_model_access_denied"
@ -273,9 +273,9 @@ async def test_team_model_access_update():
# Test updated access
async with aiohttp.ClientSession() as session:
# Both models should now work
await mock_chat_completion(session=session, key=key, model="openai/gpt-4")
await mock_chat_completion(session=session, key=key, model="openai/gpt-5.5")
await mock_chat_completion(
session=session, key=key, model="openai/gpt-3.5-turbo"
session=session, key=key, model="openai/gpt-5-mini"
)
# Non-OpenAI model should still fail

2
tests/otel_tests/test_guardrails.py
View file

@ -11,7 +11,7 @@ async def chat_completion(
session,
key,
messages,
model: Union[str, List] = "gpt-4",
model: Union[str, List] = "gpt-5.5",
guardrails: Optional[List] = None,
):
url = "http://0.0.0.0:4000/chat/completions"

6
tests/otel_tests/test_otel.py
View file

@ -11,8 +11,8 @@ from litellm._uuid import uuid
async def generate_key(
session,
models=[
"gpt-4",
"text-embedding-ada-002",
"gpt-5.5",
"text-embedding-3-small",
"gpt-image-1",
"fake-openai-endpoint",
"mistral-embed",
@ -38,7 +38,7 @@ async def generate_key(
return await response.json()
async def chat_completion(session, key, model: Union[str, List] = "gpt-4"):
async def chat_completion(session, key, model: Union[str, List] = "gpt-5.5"):
url = "http://0.0.0.0:4000/chat/completions"
headers = {
"Authorization": f"Bearer {key}",

2
tests/otel_tests/test_prometheus.py
View file

@ -177,7 +177,7 @@ async def test_proxy_failure_metrics():
@pytest.mark.flaky(retries=3, delay=2)
async def test_proxy_success_metrics():
"""
Make 1 good /chat/completions call to "openai/gpt-3.5-turbo"
Make 1 good /chat/completions call to "openai/gpt-5-mini"
GET /metrics
Assert the success metric is incremented by 1
"""

6
tests/pass_through_unit_tests/base_anthropic_messages_tool_search_test.py
View file

@ -98,9 +98,9 @@ class BaseAnthropicMessagesToolSearchTest(ABC):
Returns the model string to use for tests.
Examples:
- "anthropic/claude-sonnet-4-20250514"
- "vertex_ai/claude-sonnet-4@20250514"
- "bedrock/invoke/anthropic.claude-sonnet-4-20250514-v1:0"
- "anthropic/claude-sonnet-4-5-20250929"
- "vertex_ai/claude-sonnet-4-5@20250929"
- "bedrock/invoke/anthropic.claude-sonnet-4-5-20250929-v1:0"
"""
pass

20
tests/pass_through_unit_tests/test_anthropic_messages_passthrough.py
View file

@ -112,7 +112,7 @@ class TestAnthropicOpenAIAPI(BaseAnthropicMessagesTest):
@property
def model_config(self) -> Dict[str, Any]:
return {
"model": "openai/gpt-4o-mini",
"model": "openai/gpt-4.1-mini",
"client": None,
}
@ -121,7 +121,7 @@ class TestAnthropicOpenAIAPI(BaseAnthropicMessagesTest):
"""
This is the model name that is expected to be in the logging payload
"""
return "gpt-4o-mini"
return "gpt-4.1-mini"
@pytest.mark.asyncio
async def test_anthropic_messages_litellm_router_streaming_with_logging(self):
@ -283,23 +283,23 @@ async def test_anthropic_messages_fallbacks():
router = Router(
model_list=[
{
"model_name": "anthropic/claude-opus-4-20250514",
"model_name": "anthropic/claude-opus-4-7",
"litellm_params": {
"model": "anthropic/claude-opus-4-20250514",
"model": "anthropic/claude-opus-4-7",
"api_key": "bad-key",
},
},
{
"model_name": "bedrock/us.anthropic.claude-sonnet-4-20250514-v1:0",
"model_name": "bedrock/us.anthropic.claude-sonnet-4-5-20250929-v1:0",
"litellm_params": {
"model": "bedrock/us.anthropic.claude-sonnet-4-20250514-v1:0",
"model": "bedrock/us.anthropic.claude-sonnet-4-5-20250929-v1:0",
},
},
],
fallbacks=[
{
"anthropic/claude-opus-4-20250514": [
"bedrock/us.anthropic.claude-sonnet-4-20250514-v1:0"
"anthropic/claude-opus-4-7": [
"bedrock/us.anthropic.claude-sonnet-4-5-20250929-v1:0"
]
}
],
@ -311,7 +311,7 @@ async def test_anthropic_messages_fallbacks():
# Call the handler
response = await router.aanthropic_messages(
messages=messages,
model="anthropic/claude-opus-4-20250514",
model="anthropic/claude-opus-4-7",
max_tokens=100,
metadata={
"user_id": "hello",
@ -871,7 +871,7 @@ def test_sync_openai_messages():
litellm._turn_on_debug()
response = litellm.anthropic.messages.create(
messages=[{"role": "user", "content": "Hello, can you tell me a short joke?"}],
model="openai/gpt-4o-mini",
model="openai/gpt-4.1-mini",
max_tokens=100,
)
print("ANT response", response)

2
tests/pass_through_unit_tests/test_anthropic_messages_tool_search.py
View file

@ -50,7 +50,7 @@ class TestAnthropicAPIToolSearch(BaseAnthropicMessagesToolSearchTest):
# """
# def get_model(self) -> str:
# return "azure/claude-sonnet-4-20250514"
# return "azure/claude-sonnet-4-5-20250929"
# class TestVertexAIToolSearch(BaseAnthropicMessagesToolSearchTest):

22
tests/pass_through_unit_tests/test_bedrock_anthropic_messages_test.py
View file

@ -28,15 +28,15 @@ async def test_anthropic_messages_litellm_router_bedrock():
router = Router(
model_list=[
{
"model_name": "bedrock/converse/us.anthropic.claude-sonnet-4-20250514-v1:0",
"model_name": "bedrock/converse/us.anthropic.claude-sonnet-4-5-20250929-v1:0",
"litellm_params": {
"model": "bedrock/converse/us.anthropic.claude-sonnet-4-20250514-v1:0",
"model": "bedrock/converse/us.anthropic.claude-sonnet-4-5-20250929-v1:0",
},
},
{
"model_name": "bedrock/us.anthropic.claude-sonnet-4-20250514-v1:0",
"model_name": "bedrock/us.anthropic.claude-sonnet-4-5-20250929-v1:0",
"litellm_params": {
"model": "bedrock/us.anthropic.claude-sonnet-4-20250514-v1:0",
"model": "bedrock/us.anthropic.claude-sonnet-4-5-20250929-v1:0",
},
},
]
@ -45,20 +45,20 @@ async def test_anthropic_messages_litellm_router_bedrock():
# Set up test parameters
messages = [{"role": "user", "content": "Hello, can you tell me a short joke?"}]
# Call 1 using bedrock/converse/us.anthropic.claude-sonnet-4-20250514-v1:0
# Call 1 using bedrock/converse/us.anthropic.claude-sonnet-4-5-20250929-v1:0
response = await router.aanthropic_messages(
messages=messages,
model="bedrock/converse/us.anthropic.claude-sonnet-4-20250514-v1:0",
model="bedrock/converse/us.anthropic.claude-sonnet-4-5-20250929-v1:0",
max_tokens=100,
)
# Verify response
INSTANCE_BASE_ANTHROPIC_MESSAGES_TEST._validate_response(response)
# Call 2 using bedrock/us.anthropic.claude-sonnet-4-20250514-v1:0
# Call 2 using bedrock/us.anthropic.claude-sonnet-4-5-20250929-v1:0
response = await router.aanthropic_messages(
messages=messages,
model="bedrock/us.anthropic.claude-sonnet-4-20250514-v1:0",
model="bedrock/us.anthropic.claude-sonnet-4-5-20250929-v1:0",
max_tokens=100,
)
@ -75,9 +75,9 @@ async def test_anthropic_messages_bedrock_converse_with_thinking():
router = Router(
model_list=[
{
"model_name": "bedrock/converse/us.anthropic.claude-sonnet-4-20250514-v1:0",
"model_name": "bedrock/converse/us.anthropic.claude-sonnet-4-5-20250929-v1:0",
"litellm_params": {
"model": "bedrock/converse/us.anthropic.claude-sonnet-4-20250514-v1:0",
"model": "bedrock/converse/us.anthropic.claude-sonnet-4-5-20250929-v1:0",
},
},
]
@ -87,7 +87,7 @@ async def test_anthropic_messages_bedrock_converse_with_thinking():
response = await router.aanthropic_messages(
messages=messages,
model="bedrock/converse/us.anthropic.claude-sonnet-4-20250514-v1:0",
model="bedrock/converse/us.anthropic.claude-sonnet-4-5-20250929-v1:0",
max_tokens=1026,
thinking={"type": "enabled", "budget_tokens": 1025},
)

2
tests/pass_through_unit_tests/test_custom_logger_passthrough.py
View file

@ -45,7 +45,7 @@ async def test_assistants_passthrough_logging():
"instructions": "You are a personal math tutor. When asked a question, write and run Python code to answer the question.",
"name": "Math Tutor",
"tools": [{"type": "code_interpreter"}],
"model": "gpt-4o",
"model": "gpt-4.1-mini",
}
TARGET_METHOD = "POST"

10
tests/pass_through_unit_tests/test_pass_through_unit_tests.py
View file

@ -451,7 +451,7 @@ def test_init_kwargs_filters_pricing_params(mock_request, mock_user_api_key_dict
# Create a parsed body with pricing parameters that should be filtered out
parsed_body = {
"model": "gpt-4",
"model": "gpt-5.5",
"messages": [{"role": "user", "content": "test"}],
# Standard pricing params (should be filtered)
"input_cost_per_token": 0.00002,
@ -491,7 +491,7 @@ def test_init_kwargs_filters_pricing_params(mock_request, mock_user_api_key_dict
_parsed_body=parsed_body,
litellm_call_id="test-call-id",
logging_obj=LiteLLMLoggingObj(
model="gpt-4",
model="gpt-5.5",
messages=[{"role": "user", "content": "test"}],
stream=False,
call_type="completion",
@ -520,7 +520,7 @@ def test_init_kwargs_filters_pricing_params(mock_request, mock_user_api_key_dict
assert "tiered_pricing" not in parsed_body
# Verify valid OpenAI parameters remain in parsed_body
assert parsed_body["model"] == "gpt-4"
assert parsed_body["model"] == "gpt-5.5"
assert parsed_body["messages"] == [{"role": "user", "content": "test"}]
assert parsed_body["temperature"] == 0.7
assert parsed_body["max_tokens"] == 100
@ -560,7 +560,7 @@ def test_custom_pricing_used_in_cost_calculation():
)
],
created=1234567890,
model="gpt-4",
model="gpt-5.5",
object="chat.completion",
usage=Usage(prompt_tokens=100, completion_tokens=50, total_tokens=150),
)
@ -568,7 +568,7 @@ def test_custom_pricing_used_in_cost_calculation():
# Test 1: Standard pricing (should use default model pricing)
standard_cost = completion_cost(
completion_response=resp,
model="gpt-4",
model="gpt-5.5",
)
print(f"Standard cost: {standard_cost}")

12
tests/pass_through_unit_tests/test_unit_test_anthropic_pass_through.py
View file

@ -23,7 +23,7 @@ from litellm.proxy.pass_through_endpoints.llm_provider_handlers.anthropic_passth
@pytest.fixture
def mock_response():
return {
"model": "claude-3-opus-20240229",
"model": "claude-opus-4-7",
"content": [{"text": "Hello, world!", "type": "text"}],
"role": "assistant",
}
@ -50,7 +50,7 @@ def mock_httpx_response():
@pytest.fixture
def mock_logging_obj():
logging_obj = LiteLLMLoggingObj(
model="claude-3-opus-20240229",
model="claude-opus-4-7",
messages=[],
stream=False,
call_type="completion",
@ -101,7 +101,7 @@ def test_create_anthropic_response_logging_payload(mock_logging_obj, metadata_pa
result = AnthropicPassthroughLoggingHandler._create_anthropic_response_logging_payload(
litellm_model_response=model_response,
model="claude-3-opus-20240229",
model="claude-opus-4-7",
kwargs={
"litellm_params": {
"metadata": {
@ -249,7 +249,7 @@ def test_get_user_from_metadata(end_user_id):
def all_chunks():
return [
"event: message_start",
'data: {"type":"message_start","message":{"id":"msg_01G7T4YSBzHjmgTyizv1UfkB","type":"message","role":"assistant","model":"claude-3-5-sonnet-20240620","content":[],"stop_reason":null,"stop_sequence":null,"usage":{"input_tokens":17,"cache_creation_input_tokens":0,"cache_read_input_tokens":0,"output_tokens":5}}}',
'data: {"type":"message_start","message":{"id":"msg_01G7T4YSBzHjmgTyizv1UfkB","type":"message","role":"assistant","model":"claude-sonnet-4-5-20250929","content":[],"stop_reason":null,"stop_sequence":null,"usage":{"input_tokens":17,"cache_creation_input_tokens":0,"cache_read_input_tokens":0,"output_tokens":5}}}',
"event: content_block_start",
'data: {"type":"content_block_start","index":0,"content_block":{"type":"text","text":""}}',
"event: ping",
@ -325,7 +325,7 @@ def test_handle_logging_anthropic_collected_chunks(all_chunks):
"passthrough_success_handler_obj": pass_through_logging_obj,
"url_route": "https://api.anthropic.com/v1/messages",
"request_body": {
"model": "claude-3-5-sonnet-20240620",
"model": "claude-sonnet-4-5-20250929",
"messages": [
{
"role": "user",
@ -366,7 +366,7 @@ def test_build_complete_streaming_response(all_chunks):
result = AnthropicPassthroughLoggingHandler._build_complete_streaming_response(
all_chunks=all_chunks,
model="claude-3-5-sonnet-20240620",
model="claude-sonnet-4-5-20250929",
litellm_logging_obj=litellm_logging_obj,
)

4
tests/proxy_e2e_anthropic_messages_tests/test_config.yaml
View file

@ -9,9 +9,9 @@ model_list:
model: "bedrock/us.anthropic.claude-haiku-4-5-20251001-v1:0"
aws_region_name: "us-east-1"
- model_name: bedrock-claude-sonnet-4
- model_name: bedrock-claude-sonnet-4.6
litellm_params:
model: "bedrock/us.anthropic.claude-sonnet-4-20250514-v1:0"
model: "bedrock/us.anthropic.claude-sonnet-4-6"
aws_region_name: "us-east-1"
- model_name: bedrock-claude-sonnet-4.5

8
tests/router_unit_tests/create_mock_standard_logging_payload.py
View file

@ -43,9 +43,9 @@ def create_standard_logging_payload() -> StandardLoggingPayload:
endTime=1234567891.0,
completionStartTime=1234567890.5,
model_map_information=StandardLoggingModelInformation(
model_map_key="gpt-3.5-turbo", model_map_value=None
model_map_key="gpt-5-mini", model_map_value=None
),
model="gpt-3.5-turbo",
model="gpt-5-mini",
model_id="model-123",
model_group="openai-gpt",
api_base="https://api.openai.com",
@ -94,9 +94,9 @@ def create_standard_logging_payload_with_long_content() -> StandardLoggingPayloa
endTime=1234567891.0,
completionStartTime=1234567890.5,
model_map_information=StandardLoggingModelInformation(
model_map_key="gpt-3.5-turbo", model_map_value=None
model_map_key="gpt-5-mini", model_map_value=None
),
model="gpt-3.5-turbo",
model="gpt-5-mini",
model_id="model-123",
model_group="openai-gpt",
api_base="https://api.openai.com",

8
tests/router_unit_tests/test_completion_no_copy.py
View file

@ -28,7 +28,7 @@ async def test_acompletion_deployment_not_mutated():
{
"model_name": "gpt-3.5",
"litellm_params": {
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"api_key": "test-key",
"temperature": 0.7,
},
@ -46,7 +46,7 @@ async def test_acompletion_deployment_not_mutated():
mock_acompletion.return_value = ModelResponse(
id="test",
choices=[{"message": {"role": "assistant", "content": "test"}, "index": 0}],
model="gpt-3.5-turbo",
model="gpt-5-mini",
usage={"prompt_tokens": 10, "completion_tokens": 20, "total_tokens": 30},
)
@ -76,7 +76,7 @@ def test_completion_deployment_not_mutated():
{
"model_name": "gpt-3.5",
"litellm_params": {
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"api_key": "test-key",
"max_tokens": 100,
},
@ -94,7 +94,7 @@ def test_completion_deployment_not_mutated():
mock_completion.return_value = ModelResponse(
id="test",
choices=[{"message": {"role": "assistant", "content": "test"}, "index": 0}],
model="gpt-3.5-turbo",
model="gpt-5-mini",
usage={"prompt_tokens": 10, "completion_tokens": 20, "total_tokens": 30},
)

4
tests/router_unit_tests/test_default_deployment_copy.py
View file

@ -42,7 +42,7 @@ def test_default_deployment_isolation():
router.default_deployment = { # type: ignore
"model_name": "default-model",
"litellm_params": {
"model": "gpt-3.5-turbo", # This will be overwritten per request
"model": "gpt-5-mini", # This will be overwritten per request
"api_key": "test-key", # This should be shared
"custom_config": { # Deep nested - will be SHARED
"nested_setting": "original",
@ -66,7 +66,7 @@ def test_default_deployment_isolation():
assert deployment2["litellm_params"]["model"] == "custom-model-2" # type: ignore
# Assert: Original default_deployment must remain unchanged (not mutated by requests)
assert router.default_deployment["litellm_params"]["model"] == "gpt-3.5-turbo" # type: ignore
assert router.default_deployment["litellm_params"]["model"] == "gpt-5-mini" # type: ignore
# Assert: Shared fields should still be accessible in all copies
assert deployment1["litellm_params"]["api_key"] == "test-key" # type: ignore

18
tests/router_unit_tests/test_get_model_list_alias_optimization.py
View file

@ -10,18 +10,18 @@ def test_get_model_list_from_model_alias_should_not_iterate_for_non_alias_lookup
router = Router(
model_list=[
{
"model_name": "gpt-3.5-turbo",
"litellm_params": {"model": "gpt-3.5-turbo"},
"model_name": "gpt-5-mini",
"litellm_params": {"model": "gpt-5-mini"},
}
],
model_group_alias={"alias-1": "gpt-4"},
model_group_alias={"alias-1": "gpt-5.5"},
)
router.model_group_alias = NoItemsAliasDict(
{f"alias-{idx}": "gpt-4" for idx in range(200)}
{f"alias-{idx}": "gpt-5.5" for idx in range(200)}
)
model_alias_list = router.get_model_list_from_model_alias(
model_name="gpt-3.5-turbo"
model_name="gpt-5-mini"
)
assert model_alias_list == []
@ -30,18 +30,18 @@ def test_map_team_model_should_not_iterate_aliases_for_non_alias_team_model_name
router = Router(
model_list=[
{
"model_name": "gpt-3.5-turbo",
"litellm_params": {"model": "gpt-3.5-turbo"},
"model_name": "gpt-5-mini",
"litellm_params": {"model": "gpt-5-mini"},
"model_info": {
"team_id": "team-1",
"team_public_model_name": "team-model",
},
}
],
model_group_alias={"alias-1": "gpt-4"},
model_group_alias={"alias-1": "gpt-5.5"},
)
router.model_group_alias = NoItemsAliasDict(
{f"alias-{idx}": "gpt-4" for idx in range(200)}
{f"alias-{idx}": "gpt-5.5" for idx in range(200)}
)
# map_team_model should return the public name unchanged (not the internal UUID name)

16
tests/router_unit_tests/test_pre_call_checks_optimization.py
View file

@ -37,13 +37,13 @@ class TestPreCallChecksOptimization:
router = Router(
model_list=[
{
"model_name": "gpt-3.5-turbo",
"litellm_params": {"model": "gpt-3.5-turbo", "api_key": "sk-test"},
"model_name": "gpt-5-mini",
"litellm_params": {"model": "gpt-5-mini", "api_key": "sk-test"},
"model_info": {"id": "test-1"},
},
{
"model_name": "gpt-3.5-turbo",
"litellm_params": {"model": "gpt-4", "api_key": "sk-test2"},
"model_name": "gpt-5-mini",
"litellm_params": {"model": "gpt-5.5", "api_key": "sk-test2"},
"model_info": {"id": "test-2"},
},
],
@ -51,7 +51,7 @@ class TestPreCallChecksOptimization:
enable_pre_call_checks=True,
)
deployments = router.get_model_list(model_name="gpt-3.5-turbo")
deployments = router.get_model_list(model_name="gpt-5-mini")
assert deployments is not None
# Capture the original state
@ -62,7 +62,7 @@ class TestPreCallChecksOptimization:
# Call the function under test
router._pre_call_checks(
model="gpt-3.5-turbo",
model="gpt-5-mini",
healthy_deployments=deployments,
messages=[{"role": "user", "content": "test"}],
)
@ -92,12 +92,12 @@ class TestPreCallChecksOptimization:
model_list=[
{
"model_name": "test",
"litellm_params": {"model": "gpt-3.5-turbo", "api_key": "sk-test"},
"litellm_params": {"model": "gpt-5-mini", "api_key": "sk-test"},
"model_info": {"id": "small", "max_input_tokens": 50},
},
{
"model_name": "test",
"litellm_params": {"model": "gpt-4", "api_key": "sk-test"},
"litellm_params": {"model": "gpt-5.5", "api_key": "sk-test"},
"model_info": {"id": "large", "max_input_tokens": 10000},
},
],

10
tests/router_unit_tests/test_prompt_management_check.py
View file

@ -19,7 +19,7 @@ def test_is_prompt_management_model_optimization():
Optimization: Check if "/" in model name before calling expensive
get_model_list(). This short-circuits 99% of requests that use
standard model names like "gpt-4", "claude-3", etc.
standard model names like "gpt-5.5", "claude-3", etc.
Tests both negative (early exit) and positive (actual detection) cases.
"""
@ -29,17 +29,17 @@ def test_is_prompt_management_model_optimization():
router = Router(
model_list=[
{
"model_name": "gpt-4",
"litellm_params": {"model": "gpt-4"},
"model_name": "gpt-5.5",
"litellm_params": {"model": "gpt-5.5"},
},
{
"model_name": "claude-3",
"litellm_params": {"model": "anthropic/claude-3-sonnet-20240229"},
"litellm_params": {"model": "anthropic/claude-sonnet-4-5-20250929"},
},
]
)
assert router._is_prompt_management_model("gpt-4") is False
assert router._is_prompt_management_model("gpt-5.5") is False
assert router._is_prompt_management_model("claude-3") is False
# Test 2: Models with "/" but not in model_list -> False after check

6
tests/router_unit_tests/test_router_acancel_batch.py
View file

@ -21,9 +21,9 @@ def router():
return Router(
model_list=[
{
"model_name": "gpt-4",
"model_name": "gpt-5.5",
"litellm_params": {
"model": "gpt-4",
"model": "gpt-5.5",
"api_key": "fake-key",
},
}
@ -44,7 +44,7 @@ async def test_router_acancel_batch(router):
# This tests that the router method exists and can be called
# The actual API call is mocked
response = await router.acancel_batch(
model="gpt-4",
model="gpt-5.5",
batch_id="batch_123",
)

4
tests/router_unit_tests/test_router_batch_utils.py
View file

@ -31,11 +31,11 @@ def sample_jsonl_data() -> List[Dict]:
return [
{
"body": {
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"messages": [{"role": "user", "content": "Hello"}],
}
},
{"body": {"model": "gpt-4", "messages": [{"role": "user", "content": "Hi"}]}},
{"body": {"model": "gpt-5.5", "messages": [{"role": "user", "content": "Hi"}]}},
]

20
tests/router_unit_tests/test_router_cooldown_utils.py
View file

@ -62,8 +62,8 @@ def testing_litellm_router():
return Router(
model_list=[
{
"model_name": "gpt-3.5-turbo",
"litellm_params": {"model": "gpt-3.5-turbo"},
"model_name": "gpt-5-mini",
"litellm_params": {"model": "gpt-5-mini"},
"model_id": "test_deployment",
},
{
@ -113,7 +113,7 @@ def test_should_cooldown_deployment_rate_limit_error(testing_litellm_router):
"""
# Test 429 error (rate limit) -> always cooldown a deployment returning 429s
_exception = litellm.exceptions.RateLimitError(
"Rate limit", "openai", "gpt-3.5-turbo"
"Rate limit", "openai", "gpt-5-mini"
)
assert (
_should_cooldown_deployment(
@ -129,7 +129,7 @@ def test_should_cooldown_deployment_auth_limit_error(testing_litellm_router):
"""
# Test 401 error (auth limit) -> always cooldown a deployment returning 401s
_exception = litellm.exceptions.AuthenticationError(
"Unauthorized", "openai", "gpt-3.5-turbo"
"Unauthorized", "openai", "gpt-5-mini"
)
assert (
_should_cooldown_deployment(
@ -151,7 +151,7 @@ async def test_should_cooldown_deployment(testing_litellm_router):
# Test 429 error (rate limit) -> always cooldown a deployment returning 429s
_exception = litellm.exceptions.RateLimitError(
"Rate limit", "openai", "gpt-3.5-turbo"
"Rate limit", "openai", "gpt-5-mini"
)
assert (
_should_cooldown_deployment(
@ -211,8 +211,8 @@ async def test_should_cooldown_deployment_allowed_fails_set_on_router():
router = Router(
model_list=[
{
"model_name": "gpt-3.5-turbo",
"litellm_params": {"model": "gpt-3.5-turbo"},
"model_name": "gpt-5-mini",
"litellm_params": {"model": "gpt-5-mini"},
"model_id": "test_deployment",
},
]
@ -295,8 +295,8 @@ def router():
return Router(
model_list=[
{
"model_name": "gpt-4",
"litellm_params": {"model": "gpt-4"},
"model_name": "gpt-5.5",
"litellm_params": {"model": "gpt-5.5"},
"model_info": {
"id": "gpt-4--0",
},
@ -445,7 +445,7 @@ def test_should_cooldown_deployment_minimum_request_threshold(testing_litellm_ro
)
_exception = litellm.exceptions.InternalServerError(
"Internal error", "openai", "gpt-3.5-turbo"
"Internal error", "openai", "gpt-5-mini"
)
# With only 1 request, should NOT cooldown (below minimum threshold)

58
tests/router_unit_tests/test_router_embedding_headers.py
View file

@ -32,9 +32,9 @@ class TestRouterEmbeddingHeaders:
"""
model_list = [
{
"model_name": "text-embedding-ada-002",
"model_name": "text-embedding-3-small",
"litellm_params": {
"model": "text-embedding-ada-002",
"model": "text-embedding-3-small",
"api_key": "fake-key",
},
}
@ -53,12 +53,12 @@ class TestRouterEmbeddingHeaders:
data=[{"embedding": [0.1, 0.2, 0.3]}]
)
router.embedding(model="text-embedding-ada-002", input=["test input"])
router.embedding(model="text-embedding-3-small", input=["test input"])
# Verify _update_kwargs_before_fallbacks was called
mock_update.assert_called_once()
call_kwargs = mock_update.call_args[1]
assert call_kwargs["model"] == "text-embedding-ada-002"
assert call_kwargs["model"] == "text-embedding-3-small"
assert "kwargs" in call_kwargs
@pytest.mark.asyncio
@ -70,9 +70,9 @@ class TestRouterEmbeddingHeaders:
"""
model_list = [
{
"model_name": "text-embedding-ada-002",
"model_name": "text-embedding-3-small",
"litellm_params": {
"model": "text-embedding-ada-002",
"model": "text-embedding-3-small",
"api_key": "fake-key",
},
}
@ -94,13 +94,13 @@ class TestRouterEmbeddingHeaders:
)
await router.aembedding(
model="text-embedding-ada-002", input=["test input"]
model="text-embedding-3-small", input=["test input"]
)
# Verify _update_kwargs_before_fallbacks was called
mock_update.assert_called_once()
call_kwargs = mock_update.call_args[1]
assert call_kwargs["model"] == "text-embedding-ada-002"
assert call_kwargs["model"] == "text-embedding-3-small"
assert "kwargs" in call_kwargs
def test_embedding_propagates_default_litellm_params(self):
@ -114,9 +114,9 @@ class TestRouterEmbeddingHeaders:
model_list = [
{
"model_name": "text-embedding-ada-002",
"model_name": "text-embedding-3-small",
"litellm_params": {
"model": "text-embedding-ada-002",
"model": "text-embedding-3-small",
"api_key": "fake-key",
},
}
@ -136,7 +136,7 @@ class TestRouterEmbeddingHeaders:
data=[{"embedding": [0.1, 0.2, 0.3]}]
)
router.embedding(model="text-embedding-ada-002", input=["test input"])
router.embedding(model="text-embedding-3-small", input=["test input"])
# Verify that litellm.embedding was called with the headers
mock_litellm_embedding.assert_called_once()
@ -149,7 +149,7 @@ class TestRouterEmbeddingHeaders:
# Check that metadata was properly set up
assert "metadata" in call_kwargs
assert "model_group" in call_kwargs["metadata"]
assert call_kwargs["metadata"]["model_group"] == "text-embedding-ada-002"
assert call_kwargs["metadata"]["model_group"] == "text-embedding-3-small"
@pytest.mark.asyncio
async def test_aembedding_propagates_default_litellm_params(self):
@ -160,9 +160,9 @@ class TestRouterEmbeddingHeaders:
model_list = [
{
"model_name": "text-embedding-ada-002",
"model_name": "text-embedding-3-small",
"litellm_params": {
"model": "text-embedding-ada-002",
"model": "text-embedding-3-small",
"api_key": "fake-key",
},
}
@ -185,7 +185,7 @@ class TestRouterEmbeddingHeaders:
)
await router.aembedding(
model="text-embedding-ada-002", input=["test input"]
model="text-embedding-3-small", input=["test input"]
)
# Verify that litellm.aembedding was called with the headers
@ -199,7 +199,7 @@ class TestRouterEmbeddingHeaders:
# Check that metadata was properly set up
assert "metadata" in call_kwargs
assert "model_group" in call_kwargs["metadata"]
assert call_kwargs["metadata"]["model_group"] == "text-embedding-ada-002"
assert call_kwargs["metadata"]["model_group"] == "text-embedding-3-small"
def test_embedding_metadata_includes_model_group(self):
"""
@ -211,7 +211,7 @@ class TestRouterEmbeddingHeaders:
{
"model_name": "test-embedding-model",
"litellm_params": {
"model": "text-embedding-ada-002",
"model": "text-embedding-3-small",
"api_key": "fake-key",
},
}
@ -241,9 +241,9 @@ class TestRouterEmbeddingHeaders:
"""
model_list = [
{
"model_name": "text-embedding-ada-002",
"model_name": "text-embedding-3-small",
"litellm_params": {
"model": "text-embedding-ada-002",
"model": "text-embedding-3-small",
"api_key": "fake-key",
},
}
@ -257,7 +257,7 @@ class TestRouterEmbeddingHeaders:
data=[{"embedding": [0.1, 0.2, 0.3]}]
)
router.embedding(model="text-embedding-ada-002", input=["test input"])
router.embedding(model="text-embedding-3-small", input=["test input"])
# Verify num_retries was not set in the call (it's handled by function_with_fallbacks)
# The important thing is that it was set in kwargs before being passed to function_with_fallbacks
@ -272,9 +272,9 @@ class TestRouterEmbeddingHeaders:
"""
model_list = [
{
"model_name": "text-embedding-ada-002",
"model_name": "text-embedding-3-small",
"litellm_params": {
"model": "text-embedding-ada-002",
"model": "text-embedding-3-small",
"api_key": "fake-key",
},
}
@ -287,7 +287,7 @@ class TestRouterEmbeddingHeaders:
data=[{"embedding": [0.1, 0.2, 0.3]}]
)
router.embedding(model="text-embedding-ada-002", input=["test input"])
router.embedding(model="text-embedding-3-small", input=["test input"])
call_kwargs = mock_litellm_embedding.call_args[1]
@ -306,16 +306,16 @@ class TestRouterEmbeddingHeaders:
model_list = [
{
"model_name": "gpt-3.5-turbo",
"model_name": "gpt-5-mini",
"litellm_params": {
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"api_key": "fake-key",
},
},
{
"model_name": "text-embedding-ada-002",
"model_name": "text-embedding-3-small",
"litellm_params": {
"model": "text-embedding-ada-002",
"model": "text-embedding-3-small",
"api_key": "fake-key",
},
},
@ -330,7 +330,7 @@ class TestRouterEmbeddingHeaders:
mock_completion.return_value = MagicMock()
router.completion(
model="gpt-3.5-turbo", messages=[{"role": "user", "content": "test"}]
model="gpt-5-mini", messages=[{"role": "user", "content": "test"}]
)
completion_kwargs = mock_completion.call_args[1]
@ -341,7 +341,7 @@ class TestRouterEmbeddingHeaders:
data=[{"embedding": [0.1, 0.2, 0.3]}]
)
router.embedding(model="text-embedding-ada-002", input=["test input"])
router.embedding(model="text-embedding-3-small", input=["test input"])
embedding_kwargs = mock_embedding.call_args[1]

22
tests/router_unit_tests/test_router_embedding_integration.py
View file

@ -30,7 +30,7 @@ class TestRouterEmbeddingIntegration:
{
"model_name": "embedding-deployment-1",
"litellm_params": {
"model": "text-embedding-ada-002",
"model": "text-embedding-3-small",
"api_key": "key-1",
"headers": {"X-Deployment": "deployment-1"},
},
@ -38,7 +38,7 @@ class TestRouterEmbeddingIntegration:
{
"model_name": "embedding-deployment-2",
"litellm_params": {
"model": "text-embedding-ada-002",
"model": "text-embedding-3-small",
"api_key": "key-2",
"headers": {"X-Deployment": "deployment-2"},
},
@ -75,7 +75,7 @@ class TestRouterEmbeddingIntegration:
{
"model_name": "test-embedding",
"litellm_params": {
"model": "text-embedding-ada-002",
"model": "text-embedding-3-small",
"api_key": "test-key",
},
}
@ -117,7 +117,7 @@ class TestRouterEmbeddingIntegration:
{
"model_name": "test-embedding",
"litellm_params": {
"model": "text-embedding-ada-002",
"model": "text-embedding-3-small",
"api_key": "test-key",
},
}
@ -170,7 +170,7 @@ class TestRouterEmbeddingIntegration:
{
"model_name": "test-embedding",
"litellm_params": {
"model": "text-embedding-ada-002",
"model": "text-embedding-3-small",
"api_key": "test-key",
},
}
@ -194,7 +194,7 @@ class TestRouterEmbeddingIntegration:
{
"model_name": "test-embedding",
"litellm_params": {
"model": "text-embedding-ada-002",
"model": "text-embedding-3-small",
"api_key": "test-key",
},
}
@ -222,14 +222,14 @@ class TestRouterEmbeddingIntegration:
{
"model_name": "shared-embedding-model",
"litellm_params": {
"model": "text-embedding-ada-002",
"model": "text-embedding-3-small",
"api_key": "key-1",
},
},
{
"model_name": "shared-embedding-model",
"litellm_params": {
"model": "text-embedding-ada-002",
"model": "text-embedding-3-small",
"api_key": "key-2",
},
},
@ -264,14 +264,14 @@ class TestRouterEmbeddingIntegration:
{
"model_name": "primary-embedding",
"litellm_params": {
"model": "text-embedding-ada-002",
"model": "text-embedding-3-small",
"api_key": "primary-key",
},
},
{
"model_name": "fallback-embedding",
"litellm_params": {
"model": "text-embedding-ada-002",
"model": "text-embedding-3-small",
"api_key": "fallback-key",
},
},
@ -320,7 +320,7 @@ class TestRouterEmbeddingIntegration:
{
"model_name": "azure-embedding",
"litellm_params": {
"model": "azure/text-embedding-ada-002",
"model": "azure/text-embedding-3-small",
"api_key": "azure-key",
"api_base": "https://example.openai.azure.com",
"api_version": "2024-02-01",

28
tests/router_unit_tests/test_router_endpoints.py
View file

@ -31,23 +31,23 @@ import asyncio
def model_list():
return [
{
"model_name": "gpt-3.5-turbo",
"model_name": "gpt-5-mini",
"litellm_params": {
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"api_key": os.getenv("OPENAI_API_KEY"),
},
},
{
"model_name": "gpt-4o",
"model_name": "gpt-5.5",
"litellm_params": {
"model": "gpt-4o",
"model": "gpt-5.5",
"api_key": os.getenv("OPENAI_API_KEY"),
},
},
{
"model_name": "dall-e-3",
"model_name": "gpt-image-1",
"litellm_params": {
"model": "dall-e-3",
"model": "gpt-image-1",
"api_key": os.getenv("OPENAI_API_KEY"),
},
},
@ -59,9 +59,9 @@ def model_list():
},
},
{
"model_name": "claude-3-5-sonnet-20240620",
"model_name": "claude-sonnet-4-5-20250929",
"litellm_params": {
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"mock_response": "hi this is macintosh.",
},
},
@ -323,21 +323,21 @@ async def test_aaaaatext_completion_endpoint(model_list, sync_mode):
if sync_mode:
response = router.text_completion(
model="gpt-3.5-turbo",
model="gpt-5-mini",
prompt="Hello, how are you?",
mock_response="I'm fine, thank you!",
)
else:
## Test 1: user facing function
response = await router.atext_completion(
model="gpt-3.5-turbo",
model="gpt-5-mini",
prompt="Hello, how are you?",
mock_response="I'm fine, thank you!",
)
## Test 2: underlying function
response_2 = await router._atext_completion(
model="gpt-3.5-turbo",
model="gpt-5-mini",
prompt="Hello, how are you?",
mock_response="I'm fine, thank you!",
)
@ -359,12 +359,12 @@ async def test_router_with_empty_choices(model_list):
completion_tokens=10,
total_tokens=20,
),
model="gpt-3.5-turbo",
model="gpt-5-mini",
object="chat.completion",
created=1723081200,
).model_dump()
response = await router.acompletion(
model="gpt-3.5-turbo",
model="gpt-5-mini",
messages=[{"role": "user", "content": "Hello, how are you?"}],
mock_response=mock_response,
)
@ -1142,7 +1142,7 @@ async def test_init_containers_api_endpoints_managed_id_routes_via_generic_fallb
{
"model_name": "azure-router-model",
"litellm_params": {
"model": "azure/gpt-4",
"model": "azure/gpt-5.5",
"api_key": "fake-key",
"api_base": "https://westus.api.cognitive.microsoft.com",
},

16
tests/router_unit_tests/test_router_handle_error.py
View file

@ -33,7 +33,7 @@ async def test_send_llm_exception_alert_success():
# Create mock request kwargs
request_kwargs = {
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"messages": [{"role": "user", "content": "Hello"}],
}
@ -65,7 +65,7 @@ async def test_send_llm_exception_alert_no_logger():
# Create mock request kwargs
request_kwargs = {
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"messages": [{"role": "user", "content": "Hello"}],
}
@ -94,7 +94,7 @@ async def test_send_llm_exception_alert_when_proxy_server_request_in_kwargs():
# Create mock request kwargs
request_kwargs = {
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"messages": [{"role": "user", "content": "Hello"}],
"proxy_server_request": {},
}
@ -145,7 +145,7 @@ async def test_async_raise_no_deployment_exception():
# Call the function
result = await async_raise_no_deployment_exception(
litellm_router_instance=mock_router,
model="gpt-3.5-turbo",
model="gpt-5-mini",
parent_otel_span=None,
)
@ -153,7 +153,7 @@ async def test_async_raise_no_deployment_exception():
assert isinstance(result, RouterRateLimitError)
# Assert that the error has the correct properties
assert result.model == "gpt-3.5-turbo"
assert result.model == "gpt-5-mini"
assert result.cooldown_time == 30.0
assert result.enable_pre_call_checks is True
@ -166,7 +166,7 @@ async def test_async_raise_no_deployment_exception():
assert isinstance(item, str), f"Expected string ID, got {type(item)}: {item}"
# Verify mock calls
mock_router.get_model_ids.assert_called_once_with(model_name="gpt-3.5-turbo")
mock_router.get_model_ids.assert_called_once_with(model_name="gpt-5-mini")
mock_router.cooldown_cache.get_min_cooldown.assert_called_once_with(
model_ids=["deployment-1", "deployment-2"], parent_otel_span=None
)
@ -241,7 +241,7 @@ async def test_async_raise_no_deployment_exception_none_cooldown_list():
# After the defensive fix, this should handle None gracefully and return empty list
result = await async_raise_no_deployment_exception(
litellm_router_instance=mock_router,
model="gpt-4",
model="gpt-5.5",
parent_otel_span=None,
)
@ -249,7 +249,7 @@ async def test_async_raise_no_deployment_exception_none_cooldown_list():
assert isinstance(result, RouterRateLimitError)
# Assert that the error has the correct properties
assert result.model == "gpt-4"
assert result.model == "gpt-5.5"
assert result.cooldown_time == 45.0
assert result.enable_pre_call_checks is True

198
tests/router_unit_tests/test_router_helper_utils.py
View file

@ -21,9 +21,9 @@ from litellm.types.router import Deployment, LiteLLM_Params
def model_list():
return [
{
"model_name": "gpt-3.5-turbo",
"model_name": "gpt-5-mini",
"litellm_params": {
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"api_key": os.getenv("OPENAI_API_KEY"),
"tpm": 1000, # Add TPM limit so async method doesn't return early
"rpm": 100, # Add RPM limit so async method doesn't return early
@ -33,9 +33,9 @@ def model_list():
},
},
{
"model_name": "gpt-4o",
"model_name": "gpt-5.5",
"litellm_params": {
"model": "gpt-4o",
"model": "gpt-5.5",
"api_key": os.getenv("OPENAI_API_KEY"),
},
},
@ -64,8 +64,8 @@ def model_list():
def test_validate_fallbacks(model_list):
router = Router(model_list=model_list, fallbacks=[{"gpt-4o": "gpt-3.5-turbo"}])
router.validate_fallbacks(fallback_param=[{"gpt-4o": "gpt-3.5-turbo"}])
router = Router(model_list=model_list, fallbacks=[{"gpt-5.5": "gpt-5-mini"}])
router.validate_fallbacks(fallback_param=[{"gpt-5.5": "gpt-5-mini"}])
def test_routing_strategy_init(model_list):
@ -149,9 +149,9 @@ def test_print_deployment(model_list):
router = Router(model_list=model_list)
deployment = {
"model_name": "gpt-3.5-turbo",
"model_name": "gpt-5-mini",
"litellm_params": {
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"api_key": os.getenv("OPENAI_API_KEY"),
},
}
@ -190,7 +190,7 @@ def test_completion(model_list):
"""Test if the completion function is working correctly"""
router = Router(model_list=model_list)
response = router._completion(
model="gpt-3.5-turbo",
model="gpt-5-mini",
messages=[{"role": "user", "content": "Hello, how are you?"}],
mock_response="I'm fine, thank you!",
)
@ -224,7 +224,7 @@ async def test_router_acompletion_util(model_list):
"""Test if the underlying '_acompletion' function is working correctly"""
router = Router(model_list=model_list)
response = await router._acompletion(
model="gpt-3.5-turbo",
model="gpt-5-mini",
messages=[{"role": "user", "content": "Hello, how are you?"}],
mock_response="I'm fine, thank you!",
)
@ -236,7 +236,7 @@ async def test_router_abatch_completion_one_model_multiple_requests_util(model_l
"""Test if the 'abatch_completion_one_model_multiple_requests' function is working correctly"""
router = Router(model_list=model_list)
response = await router.abatch_completion_one_model_multiple_requests(
model="gpt-3.5-turbo",
model="gpt-5-mini",
messages=[
[{"role": "user", "content": "Hello, how are you?"}],
[{"role": "user", "content": "Hello, how are you?"}],
@ -253,7 +253,7 @@ async def test_router_schedule_acompletion(model_list):
"""Test if the 'schedule_acompletion' function is working correctly"""
router = Router(model_list=model_list)
response = await router.schedule_acompletion(
model="gpt-3.5-turbo",
model="gpt-5-mini",
messages=[{"role": "user", "content": "Hello, how are you?"}],
mock_response="I'm fine, thank you!",
priority=1,
@ -272,7 +272,7 @@ async def test_router_schedule_atext_completion(model_list):
) as mock_atext_completion:
mock_atext_completion.return_value = TextCompletionResponse()
response = await router.atext_completion(
model="gpt-3.5-turbo",
model="gpt-5-mini",
prompt="Hello, how are you?",
priority=1,
)
@ -291,9 +291,9 @@ async def test_router_schedule_factory(model_list):
) as mock_atext_completion:
mock_atext_completion.return_value = TextCompletionResponse()
response = await router._schedule_factory(
model="gpt-3.5-turbo",
model="gpt-5-mini",
args=(
"gpt-3.5-turbo",
"gpt-5-mini",
"Hello, how are you?",
),
priority=1,
@ -310,7 +310,7 @@ async def test_router_function_with_fallbacks(model_list, sync_mode):
"""Test if the router 'async_function_with_fallbacks' + 'function_with_fallbacks' are working correctly"""
router = Router(model_list=model_list)
data = {
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"messages": [{"role": "user", "content": "Hello, how are you?"}],
"mock_response": "I'm fine, thank you!",
"num_retries": 0,
@ -334,7 +334,7 @@ async def test_router_function_with_retries(model_list, sync_mode):
"""Test if the router 'async_function_with_retries' + 'function_with_retries' are working correctly"""
router = Router(model_list=model_list)
data = {
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"messages": [{"role": "user", "content": "Hello, how are you?"}],
"mock_response": "I'm fine, thank you!",
"num_retries": 0,
@ -355,7 +355,7 @@ async def test_router_make_call(model_list):
router = Router(model_list=model_list)
response = await router.make_call(
original_function=router._acompletion,
model="gpt-3.5-turbo",
model="gpt-5-mini",
messages=[{"role": "user", "content": "Hello, how are you?"}],
mock_response="I'm fine, thank you!",
)
@ -364,7 +364,7 @@ async def test_router_make_call(model_list):
## ATEXT_COMPLETION
response = await router.make_call(
original_function=router._atext_completion,
model="gpt-3.5-turbo",
model="gpt-5-mini",
prompt="Hello, how are you?",
mock_response="I'm fine, thank you!",
)
@ -373,7 +373,7 @@ async def test_router_make_call(model_list):
## AEMBEDDING
response = await router.make_call(
original_function=router._aembedding,
model="gpt-3.5-turbo",
model="gpt-5-mini",
input="Hello, how are you?",
mock_response=[0.1, 0.2, 0.3],
)
@ -394,7 +394,7 @@ def test_update_kwargs_with_deployment(model_list):
router = Router(model_list=model_list)
kwargs: dict = {"metadata": {}}
deployment = router.get_deployment_by_model_group_name(
model_group_name="gpt-3.5-turbo"
model_group_name="gpt-5-mini"
)
router._update_kwargs_with_deployment(
deployment=deployment,
@ -460,10 +460,10 @@ def test_get_fallback_model_group_from_fallbacks(model_list):
"""Test if the '_get_fallback_model_group_from_fallbacks' function is working correctly"""
router = Router(model_list=model_list)
fallback_model_group_name = router._get_fallback_model_group_from_fallbacks(
model_group="gpt-4o",
fallbacks=[{"gpt-4o": "gpt-3.5-turbo"}],
model_group="gpt-5.5",
fallbacks=[{"gpt-5.5": "gpt-5-mini"}],
)
assert fallback_model_group_name == "gpt-3.5-turbo"
assert fallback_model_group_name == "gpt-5-mini"
@pytest.mark.parametrize("sync_mode", [True, False])
@ -474,9 +474,9 @@ async def test_deployment_callback_on_success(sync_mode):
model_list = [
{
"model_name": "gpt-3.5-turbo",
"model_name": "gpt-5-mini",
"litellm_params": {
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"api_key": os.getenv("OPENAI_API_KEY"),
"rpm": 100,
},
@ -486,7 +486,7 @@ async def test_deployment_callback_on_success(sync_mode):
router = Router(model_list=model_list)
# Get the actual deployment ID that was generated
gpt_deployment = router.get_deployment_by_model_group_name(
model_group_name="gpt-3.5-turbo"
model_group_name="gpt-5-mini"
)
deployment_id = gpt_deployment["model_info"]["id"]
@ -496,14 +496,14 @@ async def test_deployment_callback_on_success(sync_mode):
kwargs = {
"litellm_params": {
"metadata": {
"model_group": "gpt-3.5-turbo",
"model_group": "gpt-5-mini",
},
"model_info": {"id": deployment_id},
},
"standard_logging_object": standard_logging_payload,
}
response = litellm.ModelResponse(
model="gpt-3.5-turbo",
model="gpt-5-mini",
usage={"total_tokens": 100},
)
if sync_mode:
@ -532,7 +532,7 @@ async def test_deployment_callback_on_failure(model_list):
kwargs = {
"litellm_params": {
"metadata": {
"model_group": "gpt-3.5-turbo",
"model_group": "gpt-5-mini",
},
"model_info": {"id": 100},
},
@ -547,7 +547,7 @@ async def test_deployment_callback_on_failure(model_list):
assert result is False
model_response = router.completion(
model="gpt-3.5-turbo",
model="gpt-5-mini",
messages=[{"role": "user", "content": "Hello, how are you?"}],
mock_response="I'm fine, thank you!",
)
@ -575,7 +575,7 @@ def test_deployment_callback_respects_cooldown_time(model_list):
kwargs = {
"exception": FakeException(),
"litellm_params": {
"metadata": {"model_group": "gpt-3.5-turbo"},
"metadata": {"model_group": "gpt-5-mini"},
"model_info": {"id": 100},
"cooldown_time": 0,
},
@ -610,7 +610,7 @@ def test_update_usage(model_list):
"""Test if the '_update_usage' function is working correctly"""
router = Router(model_list=model_list)
deployment = router.get_deployment_by_model_group_name(
model_group_name="gpt-3.5-turbo"
model_group_name="gpt-5-mini"
)
deployment_id = deployment["model_info"]["id"]
request_count = router._update_usage(
@ -635,14 +635,14 @@ def test_should_raise_content_policy_error(
"""Test if the '_should_raise_content_policy_error' function is working correctly"""
router = Router(
model_list=model_list,
default_fallbacks=["gpt-4o"] if fallback_type == "default" else None,
default_fallbacks=["gpt-5.5"] if fallback_type == "default" else None,
)
assert (
router._should_raise_content_policy_error(
model="gpt-3.5-turbo",
model="gpt-5-mini",
response=litellm.ModelResponse(
model="gpt-3.5-turbo",
model="gpt-5-mini",
choices=[
{
"finish_reason": finish_reason,
@ -653,7 +653,7 @@ def test_should_raise_content_policy_error(
),
kwargs={
"content_policy_fallbacks": (
[{"gpt-3.5-turbo": "gpt-4o"}]
[{"gpt-5-mini": "gpt-5.5"}]
if fallback_type == "model-specific"
else None
)
@ -667,7 +667,7 @@ def test_get_healthy_deployments(model_list):
"""Test if the '_get_healthy_deployments' function is working correctly"""
router = Router(model_list=model_list)
deployments = router._get_healthy_deployments(
model="gpt-3.5-turbo", parent_otel_span=None
model="gpt-5-mini", parent_otel_span=None
)
assert len(deployments) > 0
@ -685,11 +685,11 @@ async def test_routing_strategy_pre_call_checks(model_list, sync_mode):
router = Router(model_list=model_list)
deployment = router.get_deployment_by_model_group_name(
model_group_name="gpt-3.5-turbo"
model_group_name="gpt-5-mini"
)
litellm_logging_obj = Logging(
model="gpt-3.5-turbo",
model="gpt-5-mini",
messages=[{"role": "user", "content": "hi"}],
stream=False,
call_type="acompletion",
@ -713,7 +713,7 @@ async def test_routing_strategy_pre_call_checks(model_list, sync_mode):
side_effect=litellm.RateLimitError(
message="Rate limit error",
llm_provider="openai",
model="gpt-3.5-turbo",
model="gpt-5-mini",
)
),
):
@ -752,9 +752,9 @@ def test_create_deployment(
os.environ["LITELLM_ENVIRONMENT"] = "staging"
deployment = router._create_deployment(
deployment_info={},
_model_name="gpt-3.5-turbo",
_model_name="gpt-5-mini",
_litellm_params={
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"api_key": "test",
"custom_llm_provider": "openai",
},
@ -779,7 +779,7 @@ def test_deployment_is_active_for_environment(
"""Test if the '_deployment_is_active_for_environment' function is working correctly"""
router = Router(model_list=model_list)
deployment = router.get_deployment_by_model_group_name(
model_group_name="gpt-3.5-turbo"
model_group_name="gpt-5-mini"
)
if set_supported_environments:
os.environ["LITELLM_ENVIRONMENT"] = "staging"
@ -805,7 +805,7 @@ def test_add_deployment(model_list):
"""Test if the '_add_deployment' function is working correctly"""
router = Router(model_list=model_list)
deployment = router.get_deployment_by_model_group_name(
model_group_name="gpt-3.5-turbo"
model_group_name="gpt-5-mini"
)
deployment["model_info"]["id"] = "100"
## Test 1: call user facing function
@ -821,9 +821,9 @@ def test_upsert_deployment(model_list):
router = Router(model_list=model_list)
print("model list", len(router.model_list))
deployment = router.get_deployment_by_model_group_name(
model_group_name="gpt-3.5-turbo"
model_group_name="gpt-5-mini"
)
deployment.litellm_params.model = "gpt-4o"
deployment.litellm_params.model = "gpt-5.5"
router.upsert_deployment(deployment=deployment)
assert len(router.model_list) == len(model_list)
@ -832,7 +832,7 @@ def test_delete_deployment(model_list):
"""Test if the 'delete_deployment' function is working correctly"""
router = Router(model_list=model_list)
deployment = router.get_deployment_by_model_group_name(
model_group_name="gpt-3.5-turbo"
model_group_name="gpt-5-mini"
)
router.delete_deployment(id=deployment["model_info"]["id"])
assert len(router.model_list) == len(model_list) - 1
@ -842,7 +842,7 @@ def test_get_model_info(model_list):
"""Test if the 'get_model_info' function is working correctly"""
router = Router(model_list=model_list)
deployment = router.get_deployment_by_model_group_name(
model_group_name="gpt-3.5-turbo"
model_group_name="gpt-5-mini"
)
model_info = router.get_model_info(id=deployment["model_info"]["id"])
assert model_info is not None
@ -852,19 +852,19 @@ def test_get_model_group(model_list):
"""Test if the 'get_model_group' function is working correctly"""
router = Router(model_list=model_list)
deployment = router.get_deployment_by_model_group_name(
model_group_name="gpt-3.5-turbo"
model_group_name="gpt-5-mini"
)
model_group = router.get_model_group(id=deployment["model_info"]["id"])
assert model_group is not None
assert model_group[0]["model_name"] == "gpt-3.5-turbo"
assert model_group[0]["model_name"] == "gpt-5-mini"
@pytest.mark.parametrize("user_facing_model_group_name", ["gpt-3.5-turbo", "gpt-4o"])
@pytest.mark.parametrize("user_facing_model_group_name", ["gpt-5-mini", "gpt-5.5"])
def test_set_model_group_info(model_list, user_facing_model_group_name):
"""Test if the 'set_model_group_info' function is working correctly"""
router = Router(model_list=model_list)
resp = router._set_model_group_info(
model_group="gpt-3.5-turbo",
model_group="gpt-5-mini",
user_facing_model_group_name=user_facing_model_group_name,
)
assert resp is not None
@ -956,7 +956,7 @@ def test_get_all_deployments(model_list):
"""Test if the 'get_all_deployments' function is working correctly"""
router = Router(model_list=model_list)
deployments = router._get_all_deployments(
model_name="gpt-3.5-turbo", model_alias="gpt-3.5-turbo"
model_name="gpt-5-mini", model_alias="gpt-5-mini"
)
assert len(deployments) > 0
@ -981,7 +981,7 @@ def test_common_checks_available_deployment(model_list):
"""Test if the 'common_checks_available_deployment' function is working correctly"""
router = Router(model_list=model_list)
_, available_deployments = router._common_checks_available_deployment(
model="gpt-3.5-turbo",
model="gpt-5-mini",
messages=[{"role": "user", "content": "hi"}],
input="hi",
specific_deployment=False,
@ -994,11 +994,11 @@ def test_filter_cooldown_deployments(model_list):
"""Test if the 'filter_cooldown_deployments' function is working correctly"""
router = Router(model_list=model_list)
deployments = router._filter_cooldown_deployments(
healthy_deployments=router._get_all_deployments(model_name="gpt-3.5-turbo"), # type: ignore
healthy_deployments=router._get_all_deployments(model_name="gpt-5-mini"), # type: ignore
cooldown_deployments=[],
)
assert len(deployments) == len(
router._get_all_deployments(model_name="gpt-3.5-turbo")
router._get_all_deployments(model_name="gpt-5-mini")
)
@ -1009,10 +1009,10 @@ def test_track_deployment_metrics(model_list):
router = Router(model_list=model_list)
router._track_deployment_metrics(
deployment=router.get_deployment_by_model_group_name(
model_group_name="gpt-3.5-turbo"
model_group_name="gpt-5-mini"
),
response=ModelResponse(
model="gpt-3.5-turbo",
model="gpt-5-mini",
usage={"total_tokens": 100},
),
parent_otel_span=None,
@ -1047,7 +1047,7 @@ def test_get_num_retries_from_retry_policy(
print("exception_type", exception_type)
calc_num_retries = router.get_num_retries_from_retry_policy(
exception=exception_type(
message="test", llm_provider="openai", model="gpt-3.5-turbo"
message="test", llm_provider="openai", model="gpt-5-mini"
)
)
assert calc_num_retries == num_retries
@ -1078,7 +1078,7 @@ def test_get_allowed_fails_from_policy(
)
calc_allowed_fails = router.get_allowed_fails_from_policy(
exception=exception_type(
message="test", llm_provider="openai", model="gpt-3.5-turbo"
message="test", llm_provider="openai", model="gpt-5-mini"
)
)
assert calc_allowed_fails == allowed_fails
@ -1170,16 +1170,16 @@ def test_get_model_from_alias(model_list):
"""Test if the 'get_model_from_alias' function is working correctly"""
router = Router(
model_list=model_list,
model_group_alias={"gpt-4o": "gpt-3.5-turbo"},
model_group_alias={"gpt-5.5": "gpt-5-mini"},
)
model = router._get_model_from_alias(model="gpt-4o")
assert model == "gpt-3.5-turbo"
model = router._get_model_from_alias(model="gpt-5.5")
assert model == "gpt-5-mini"
def test_get_deployment_by_litellm_model(model_list):
"""Test if the 'get_deployment_by_litellm_model' function is working correctly"""
router = Router(model_list=model_list)
deployment = router._get_deployment_by_litellm_model(model="gpt-3.5-turbo")
deployment = router._get_deployment_by_litellm_model(model="gpt-5-mini")
assert deployment is not None
@ -1239,8 +1239,8 @@ def test_replace_model_in_jsonl(model_list):
(
"fo::hi::static::hello",
"fo::*::static::*",
"openai/gpt-3.5-turbo",
"openai/gpt-3.5-turbo",
"openai/gpt-5-mini",
"openai/gpt-5-mini",
),
(
"bedrock/meta.llama3-70b",
@ -1333,10 +1333,10 @@ async def test_async_callback_filter_deployments(model_list):
router = Router(model_list=model_list)
healthy_deployments = router.get_model_list(model_name="gpt-3.5-turbo")
healthy_deployments = router.get_model_list(model_name="gpt-5-mini")
new_healthy_deployments = await router.async_callback_filter_deployments(
model="gpt-3.5-turbo",
model="gpt-5-mini",
healthy_deployments=healthy_deployments,
messages=[],
parent_otel_span=None,
@ -1350,10 +1350,10 @@ def test_cached_get_model_group_info(model_list):
router = Router(model_list=model_list)
# First call - should hit the actual function
result1 = router._cached_get_model_group_info("gpt-3.5-turbo")
result1 = router._cached_get_model_group_info("gpt-5-mini")
# Second call with same argument - should hit the cache
result2 = router._cached_get_model_group_info("gpt-3.5-turbo")
result2 = router._cached_get_model_group_info("gpt-5-mini")
# Verify results are the same
assert result1 == result2
@ -1437,7 +1437,7 @@ def test_is_auto_router_deployment(model_list):
assert router._is_auto_router_deployment(litellm_params_auto) is True
# Test case 2: Model doesn't start with "auto_router/" - should return False
litellm_params_regular = LiteLLM_Params(model="gpt-3.5-turbo")
litellm_params_regular = LiteLLM_Params(model="gpt-5-mini")
assert router._is_auto_router_deployment(litellm_params_regular) is False
# Test case 3: Model is empty string - should return False
@ -1462,8 +1462,8 @@ def test_init_auto_router_deployment_success(mock_auto_router, model_list):
litellm_params = LiteLLM_Params(
model="auto_router/test",
auto_router_config_path="/path/to/config",
auto_router_default_model="gpt-3.5-turbo",
auto_router_embedding_model="text-embedding-ada-002",
auto_router_default_model="gpt-5-mini",
auto_router_embedding_model="text-embedding-3-small",
)
deployment = Deployment(
model_name="test-auto-router",
@ -1479,8 +1479,8 @@ def test_init_auto_router_deployment_success(mock_auto_router, model_list):
model_name="test-auto-router",
auto_router_config_path="/path/to/config",
auto_router_config=None,
default_model="gpt-3.5-turbo",
embedding_model="text-embedding-ada-002",
default_model="gpt-5-mini",
embedding_model="text-embedding-3-small",
litellm_router_instance=router,
)
@ -1505,8 +1505,8 @@ def test_init_auto_router_deployment_duplicate_model_name(mock_auto_router, mode
litellm_params = LiteLLM_Params(
model="auto_router/test",
auto_router_config_path="/path/to/config",
auto_router_default_model="gpt-3.5-turbo",
auto_router_embedding_model="text-embedding-ada-002",
auto_router_default_model="gpt-5-mini",
auto_router_embedding_model="text-embedding-3-small",
)
deployment = Deployment(
model_name="test-auto-router",
@ -1971,7 +1971,7 @@ def test_get_metadata_variable_name_from_kwargs(model_list):
# Test case 4: kwargs contains other keys but no metadata keys - should return "metadata"
kwargs_other = {
"model": "gpt-4",
"model": "gpt-5.5",
"messages": [{"role": "user", "content": "hello"}],
}
result = router._get_metadata_variable_name_from_kwargs(kwargs_other)
@ -2167,15 +2167,15 @@ def test_get_first_default_fallback():
# Test with default fallback ("*")
model_list = [
{
"model_name": "gpt-3.5-turbo",
"litellm_params": {"model": "gpt-3.5-turbo", "api_key": "fake-key"},
"model_name": "gpt-5-mini",
"litellm_params": {"model": "gpt-5-mini", "api_key": "fake-key"},
}
]
router = Router(model_list=model_list, fallbacks=[{"*": ["gpt-3.5-turbo"]}])
router = Router(model_list=model_list, fallbacks=[{"*": ["gpt-5-mini"]}])
result = router._get_first_default_fallback()
assert result == "gpt-3.5-turbo"
assert result == "gpt-5-mini"
# Test with no fallbacks
router_no_fallbacks = Router(model_list=model_list)
@ -2184,7 +2184,7 @@ def test_get_first_default_fallback():
# Test with fallbacks but no default
router_no_default = Router(
model_list=model_list, fallbacks=[{"gpt-4": ["gpt-3.5-turbo"]}]
model_list=model_list, fallbacks=[{"gpt-5.5": ["gpt-5-mini"]}]
)
result = router_no_default._get_first_default_fallback()
assert result is None
@ -2206,16 +2206,16 @@ def test_resolve_model_name_from_model_id():
# Test case 2: model_id directly matches a model_name
model_list = [
{
"model_name": "gpt-3.5-turbo",
"model_name": "gpt-5-mini",
"litellm_params": {
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"api_key": "test-key",
},
},
]
router = Router(model_list=model_list)
result = router.resolve_model_name_from_model_id("gpt-3.5-turbo")
assert result == "gpt-3.5-turbo"
result = router.resolve_model_name_from_model_id("gpt-5-mini")
assert result == "gpt-5-mini"
# Test case 3: model_id matches litellm_params.model exactly
model_list = [
@ -2268,9 +2268,9 @@ def test_resolve_model_name_from_model_id():
# Test case 6: model_id doesn't match anything
model_list = [
{
"model_name": "gpt-3.5-turbo",
"model_name": "gpt-5-mini",
"litellm_params": {
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"api_key": "test-key",
},
},
@ -2287,9 +2287,9 @@ def test_resolve_model_name_from_model_id():
# Test case 8: Multiple models, find the correct one
model_list = [
{
"model_name": "gpt-3.5-turbo",
"model_name": "gpt-5-mini",
"litellm_params": {
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"api_key": "test-key",
},
},
@ -2309,17 +2309,17 @@ def test_resolve_model_name_from_model_id():
# This tests the has_model_id path in Strategy 1
model_list = [
{
"model_name": "gpt-3.5-turbo",
"model_name": "gpt-5-mini",
"litellm_params": {
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"api_key": "test-key",
},
},
]
router = Router(model_list=model_list)
result = router.resolve_model_name_from_model_id("gpt-3.5-turbo")
assert result == "gpt-3.5-turbo"
result = router.resolve_model_name_from_model_id("gpt-5-mini")
assert result == "gpt-5-mini"
def test_get_valid_args():
@ -2356,8 +2356,8 @@ def test_get_router_model_info_with_deployment_object():
router = Router(
model_list=[
{
"model_name": "gpt-4",
"litellm_params": {"model": "gpt-4", "api_key": "test-key"},
"model_name": "gpt-5.5",
"litellm_params": {"model": "gpt-5.5", "api_key": "test-key"},
"model_info": {"id": "test-id"},
}
]
@ -2373,7 +2373,7 @@ def test_get_router_model_info_with_deployment_object():
# that reuses the existing LiteLLM_Params instead of reconstructing it
model_info = router.get_router_model_info(
deployment=deployment,
received_model_name="gpt-4",
received_model_name="gpt-5.5",
)
# Verify we got valid model info back

60
tests/router_unit_tests/test_router_index_management.py
View file

@ -22,8 +22,8 @@ class TestRouterIndexManagement:
"""Test that deleting a deployment updates model_name_to_deployment_indices correctly"""
router.model_list = [
{"model_name": "gpt-3.5", "model_info": {"id": "model-1"}},
{"model_name": "gpt-4", "model_info": {"id": "model-2"}},
{"model_name": "gpt-4", "model_info": {"id": "model-3"}},
{"model_name": "gpt-5.5", "model_info": {"id": "model-2"}},
{"model_name": "gpt-5.5", "model_info": {"id": "model-3"}},
{"model_name": "claude", "model_info": {"id": "model-4"}},
]
router.model_id_to_deployment_index_map = {
@ -34,31 +34,31 @@ class TestRouterIndexManagement:
}
router.model_name_to_deployment_indices = {
"gpt-3.5": [0],
"gpt-4": [1, 2],
"gpt-5.5": [1, 2],
"claude": [3],
}
# Remove one of the duplicate gpt-4 deployments
# Remove one of the duplicate gpt-5.5 deployments
router._update_deployment_indices_after_removal(
model_id="model-2", removal_idx=1
)
# Verify indices are shifted correctly
assert router.model_name_to_deployment_indices["gpt-3.5"] == [0]
assert router.model_name_to_deployment_indices["gpt-4"] == [
assert router.model_name_to_deployment_indices["gpt-5.5"] == [
1
] # was [1,2], removed 1, shifted 2->1
assert router.model_name_to_deployment_indices["claude"] == [
2
] # was [3], shifted to [2]
# Remove the last gpt-4 deployment
# Remove the last gpt-5.5 deployment
router._update_deployment_indices_after_removal(
model_id="model-3", removal_idx=1
)
# Verify gpt-4 is removed from dict when no deployments remain
assert "gpt-4" not in router.model_name_to_deployment_indices
# Verify gpt-5.5 is removed from dict when no deployments remain
assert "gpt-5.5" not in router.model_name_to_deployment_indices
assert router.model_name_to_deployment_indices["gpt-3.5"] == [0]
assert router.model_name_to_deployment_indices["claude"] == [1]
@ -66,13 +66,13 @@ class TestRouterIndexManagement:
"""Test _build_model_id_to_deployment_index_map function"""
model_list = [
{
"model_name": "gpt-3.5-turbo",
"litellm_params": {"model": "gpt-3.5-turbo"},
"model_name": "gpt-5-mini",
"litellm_params": {"model": "gpt-5-mini"},
"model_info": {"id": "model-1"},
},
{
"model_name": "gpt-4",
"litellm_params": {"model": "gpt-4"},
"model_name": "gpt-5.5",
"litellm_params": {"model": "gpt-5.5"},
"model_info": {"id": "model-2"},
},
]
@ -136,19 +136,19 @@ class TestRouterIndexManagement:
"model_info": {
"id": "dep-1",
"team_id": "team-abc",
"team_public_model_name": "gpt-4o",
"team_public_model_name": "gpt-5.5",
},
}
router._update_team_model_index(model, 0)
assert router.team_model_to_deployment_indices[("team-abc", "gpt-4o")] == [0]
assert router.team_model_to_deployment_indices[("team-abc", "gpt-5.5")] == [0]
router._update_team_model_index(model, 2)
assert router.team_model_to_deployment_indices[("team-abc", "gpt-4o")] == [0, 2]
assert router.team_model_to_deployment_indices[("team-abc", "gpt-5.5")] == [0, 2]
router._update_team_model_index(
{"model_name": "x", "model_info": {"id": "dep-2"}}, 5
)
assert router.team_model_to_deployment_indices == {
("team-abc", "gpt-4o"): [0, 2],
("team-abc", "gpt-5.5"): [0, 2],
}
def test_has_model_id(self, router):
@ -183,18 +183,18 @@ class TestRouterIndexManagement:
"""Test _build_model_name_index function"""
model_list = [
{
"model_name": "gpt-3.5-turbo",
"litellm_params": {"model": "gpt-3.5-turbo"},
"model_name": "gpt-5-mini",
"litellm_params": {"model": "gpt-5-mini"},
"model_info": {"id": "model-1"},
},
{
"model_name": "gpt-4",
"litellm_params": {"model": "gpt-4"},
"model_name": "gpt-5.5",
"litellm_params": {"model": "gpt-5.5"},
"model_info": {"id": "model-2"},
},
{
"model_name": "gpt-4", # Duplicate model_name, different deployment
"litellm_params": {"model": "gpt-4"},
"model_name": "gpt-5.5", # Duplicate model_name, different deployment
"litellm_params": {"model": "gpt-5.5"},
"model_info": {"id": "model-3"},
},
]
@ -203,14 +203,14 @@ class TestRouterIndexManagement:
router._build_model_name_index(model_list)
# Verify: model_name_to_deployment_indices is correctly built
assert "gpt-3.5-turbo" in router.model_name_to_deployment_indices
assert "gpt-4" in router.model_name_to_deployment_indices
assert "gpt-5-mini" in router.model_name_to_deployment_indices
assert "gpt-5.5" in router.model_name_to_deployment_indices
# Verify: gpt-3.5-turbo has single deployment
assert router.model_name_to_deployment_indices["gpt-3.5-turbo"] == [0]
# Verify: gpt-5-mini has single deployment
assert router.model_name_to_deployment_indices["gpt-5-mini"] == [0]
# Verify: gpt-4 has multiple deployments
assert router.model_name_to_deployment_indices["gpt-4"] == [1, 2]
# Verify: gpt-5.5 has multiple deployments
assert router.model_name_to_deployment_indices["gpt-5.5"] == [1, 2]
# Test: Rebuild index (should clear and rebuild)
new_model_list = [
@ -223,8 +223,8 @@ class TestRouterIndexManagement:
router._build_model_name_index(new_model_list)
# Verify: Old entries are cleared
assert "gpt-3.5-turbo" not in router.model_name_to_deployment_indices
assert "gpt-4" not in router.model_name_to_deployment_indices
assert "gpt-5-mini" not in router.model_name_to_deployment_indices
assert "gpt-5.5" not in router.model_name_to_deployment_indices
# Verify: New entry is added
assert "claude-3" in router.model_name_to_deployment_indices

2
tests/router_unit_tests/test_router_prompt_caching.py
View file

@ -124,7 +124,7 @@ async def test_router_prompt_caching_same_cacheable_prefix_routes_to_same_deploy
{
"model_name": "test-model",
"litellm_params": {
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"api_base": "https://exampleopenaiendpoint-production-0ee2.up.railway.app/v1",
"api_key": f"test-key-{i}",
},

4
tests/spend_tracking_tests/test_ocr_spend_tracking.py
View file

@ -231,7 +231,7 @@ class TestGetLoggingPayloadOCR:
def test_non_ocr_call_uses_token_based_usage(self, mock_datetime):
"""Test that non-OCR calls still use token-based usage"""
kwargs = {
"model": "gpt-4",
"model": "gpt-5.5",
"call_type": "completion",
"litellm_params": {},
"response_cost": 0.02,
@ -240,7 +240,7 @@ class TestGetLoggingPayloadOCR:
response_obj = {
"id": "completion-test-123",
"object": "chat.completion",
"model": "gpt-4",
"model": "gpt-5.5",
"usage": {
"prompt_tokens": 50,
"completion_tokens": 100,

2
tests/spend_tracking_tests/test_spend_accuracy_tests.py
View file

@ -38,7 +38,7 @@ Additional Test Scenarios:
# Upstream model the proxy is configured with (spend_tracking_config.yaml).
# The proxy computes spend using this model's pricing; the local ground-truth
# calculation uses the same pricing table via litellm.cost_per_token.
UPSTREAM_MODEL = "gpt-3.5-turbo"
UPSTREAM_MODEL = "gpt-5-mini"
# Batch writer flush cadence in CI is ~2-7s (PROXY_BATCH_WRITE_AT=2 + up to 5s jitter).
# Poll every 2s for 60s — plenty of headroom for multiple ticks to land.

4
tests/test_end_users.py
View file

@ -180,7 +180,7 @@ async def test_aaaend_user_specific_region():
## MAKE CALL ##
key_gen = await generate_key(
session=session, i=0, models=["gpt-3.5-turbo-end-user-test"]
session=session, i=0, models=["gpt-5-mini-end-user-test"]
)
key = key_gen["key"]
@ -190,7 +190,7 @@ async def test_aaaend_user_specific_region():
print("SENDING USER PARAM - {}".format(end_user_obj["user_id"]))
result = await client.chat.completions.with_raw_response.create(
model="gpt-3.5-turbo-end-user-test",
model="gpt-5-mini-end-user-test",
messages=[{"role": "user", "content": "Hey!"}],
user=end_user_obj["user_id"],
)

123
tests/test_litellm/proxy/test_proxy_server.py
View file

@ -3850,6 +3850,65 @@ def test_update_config_fields_uppercases_env_vars(monkeypatch):
assert os.environ.get("DD_SITE") == "us5.datadoghq.com"
def test_encrypt_env_variables_for_db_is_idempotent(monkeypatch):
"""
Regression: /config/update and save_config must not stack a second
encryption layer when a caller re-submits a value that is already
ciphertext (the Admin UI reads config back from /get/config/callbacks —
which returns the stored, still-encrypted value — and re-POSTs it on the
next save). _encrypt_env_variables_for_db must yield a value that decrypts
to the original plaintext in exactly ONE layer, no matter how many times
its own output is fed back in. It must also not mutate os.environ (write
path — loading into the process env is the read path's job).
"""
from litellm.proxy.common_utils.encrypt_decrypt_utils import (
decrypt_value_helper,
)
from litellm.proxy.proxy_server import ProxyConfig
monkeypatch.setenv("LITELLM_SALT_KEY", "sk-test-salt-key")
monkeypatch.delenv("LANGFUSE_PUBLIC_KEY", raising=False)
proxy_config = ProxyConfig()
plaintext = "pk-langfuse-secret-value"
# First write: plaintext in -> single-encrypted out.
enc1 = proxy_config._encrypt_env_variables_for_db(
{"LANGFUSE_PUBLIC_KEY": plaintext}
)
assert enc1["LANGFUSE_PUBLIC_KEY"] != plaintext
assert (
decrypt_value_helper(
value=enc1["LANGFUSE_PUBLIC_KEY"], key="LANGFUSE_PUBLIC_KEY"
)
== plaintext
)
# UI round-trip: feed the ciphertext back in. Must NOT double-encrypt.
enc2 = proxy_config._encrypt_env_variables_for_db(enc1)
assert (
decrypt_value_helper(
value=enc2["LANGFUSE_PUBLIC_KEY"], key="LANGFUSE_PUBLIC_KEY"
)
== plaintext
)
# And again, ×3 total ciphertext re-feeds — still exactly one layer,
# never stacked, no matter how many times the UI re-saves.
enc3 = proxy_config._encrypt_env_variables_for_db(enc2)
enc4 = proxy_config._encrypt_env_variables_for_db(enc3)
for stacked in (enc3, enc4):
assert (
decrypt_value_helper(
value=stacked["LANGFUSE_PUBLIC_KEY"], key="LANGFUSE_PUBLIC_KEY"
)
== plaintext
)
# Write path must not leak the value into the process environment.
assert os.environ.get("LANGFUSE_PUBLIC_KEY") is None
def test_get_prompt_spec_for_db_prompt_with_versions():
"""
Test that _get_prompt_spec_for_db_prompt correctly converts database prompts
@ -6229,6 +6288,70 @@ def test_update_config_writes_only_sent_section(_update_config_setup):
restore()
def test_update_config_env_var_round_trip_not_double_encrypted(
_update_config_setup, monkeypatch
):
"""Endpoint-level regression for the /config/update double-encryption bug.
The Admin UI reads config back via /get/config/callbacks (which returns
the stored, still-encrypted value) and re-POSTs it on the next save. The
handler must NOT stack a second encryption layer on the re-submitted
ciphertext, and must leave untouched keys byte-identical.
Uses an invertible fake encrypt/decrypt pair ("enc:" prefix) so the
decrypt-then-encrypt chokepoint round-trips faithfully. On the pre-fix
code this stored "enc:enc:..."; the assertions below would fail there.
"""
def _fake_decrypt(
value, key=None, exception_type="error", return_original_value=False
):
if isinstance(value, str) and value.startswith("enc:"):
return value[len("enc:") :]
return value if return_original_value else None
monkeypatch.setattr(
"litellm.proxy.proxy_server.decrypt_value_helper", _fake_decrypt
)
client, prisma, restore = _update_config_setup(
initial_rows={"environment_variables": {"PREEXISTING_KEY": "enc:keepme"}}
)
try:
# First write: plaintext in -> single-encrypted at rest.
resp = client.post(
"/config/update",
json={"environment_variables": {"LANGFUSE_SECRET_KEY": "sk-secret"}},
)
assert resp.status_code == 200
stored = prisma.db.litellm_config.rows["environment_variables"]
assert stored["LANGFUSE_SECRET_KEY"] == "enc:sk-secret"
# UI round-trip: re-POST the stored ciphertext (no field change).
resp = client.post(
"/config/update",
json={
"environment_variables": {
"LANGFUSE_SECRET_KEY": stored["LANGFUSE_SECRET_KEY"]
}
},
)
assert resp.status_code == 200
stored = prisma.db.litellm_config.rows["environment_variables"]
# The bug: this would be "enc:enc:sk-secret". The fix keeps it single.
assert stored["LANGFUSE_SECRET_KEY"] == "enc:sk-secret"
assert (
_fake_decrypt(stored["LANGFUSE_SECRET_KEY"], return_original_value=True)
== "sk-secret"
)
# Untouched key preserved byte-for-byte (only sent keys rewritten).
assert stored["PREEXISTING_KEY"] == "enc:keepme"
finally:
restore()
def test_update_config_can_flip_store_model_in_db_when_currently_false(
_update_config_setup,
):

2
tests/test_litellm_proxy_responses_config.py
View file

@ -15,7 +15,7 @@ def test_litellm_proxy_responses_api_config():
)
config = ProviderConfigManager.get_provider_responses_api_config(
model="litellm_proxy/gpt-4",
model="litellm_proxy/gpt-5.5",
provider=LlmProviders.LITELLM_PROXY,
)
print(f"config: {config}")

8
tests/test_ratelimit.py
View file

@ -20,9 +20,9 @@ from litellm import utils, Router
COMPLETION_TOKENS = 5
base_model_list = [
{
"model_name": "gpt-3.5-turbo",
"model_name": "gpt-5-mini",
"litellm_params": {
"model": "gpt-3.5-turbo",
"model": "gpt-5-mini",
"api_key": os.getenv("OPENAI_API_KEY"),
"max_tokens": COMPLETION_TOKENS,
},
@ -74,14 +74,14 @@ def calculate_limits(list_of_messages):
async def async_call(router: Router, list_of_messages) -> Any:
tasks = [
router.acompletion(model="gpt-3.5-turbo", messages=m) for m in list_of_messages
router.acompletion(model="gpt-5-mini", messages=m) for m in list_of_messages
]
return await asyncio.gather(*tasks)
def sync_call(router: Router, list_of_messages) -> Any:
return [
router.completion(model="gpt-3.5-turbo", messages=m) for m in list_of_messages
router.completion(model="gpt-5-mini", messages=m) for m in list_of_messages
]

4
tests/unified_google_tests/test_litellm_responses_bridge.py
View file

@ -19,9 +19,9 @@ class TestLiteLLMResponsesBridge(BaseInteractionsTest):
"""Return the model string for the bridge provider.
The bridge provider uses litellm.responses() internally, so we can
use any model that litellm.responses() supports (e.g., gpt-4o).
use any model that litellm.responses() supports (e.g., gpt-5.5).
"""
return "gpt-4o"
return "gpt-5.5"
def get_api_key(self) -> str:
"""Return the OpenAI API key from environment."""