[Bug]: LiteLLM does not authenticate Gemini with API key. Asking for Vertex credentials.

[sanctimon]

What happened?

The LiteLLM proxy, when configured with a public Google AI Studio API key, defaults to using Google's Vertex AI endpoint and expects Application Default Credentials (ADC), causing a DefaultCredentialsError. This occurs even when the provider: gemini is explicitly set in the config.yaml.

Steps to reproduce the error:

Use a standard GEMINI_API_KEY from Google AI Studio.

Configure LiteLLM using a litellm_config.yaml with the provider: gemini and api_key: os.environ/GEMINI_API_KEY settings.

Attempt to make a chat completion request to a Gemini model (e.g., gemini-2.5-flash).

The request fails with a DefaultCredentialsError, as shown in the logs.

This issue prevents users from using a simple API key with the new LiteLLM versions, forcing them into a more complex Vertex AI authentication flow.

# litellm_config.yaml
model_list:
  - model_name: gemini-2.5-flash
    litellm_params:
      model: gemini-2.5-flash
      api_base: https://generativelanguage.googleapis.com/
      api_key: os.environ/GEMINI_API_KEY
      provider: gemini
  - model_name: gemini-2.5-pro
    litellm_params:
      model: gemini-2.5-pro
      api_base: https://generativelanguage.googleapis.com/
      api_key: os.environ/GEMINI_API_KEY
      provider: gemini

litellm_settings:
  drop_params: True
#  check_provider_endpoint: true

Relevant log output

litellm  | 23:52:19 - LiteLLM Router:INFO: router.py:1355 - litellm.acompletion(model=gemini-2.5-flash) Exception litellm.APIConnectionError: Your default credentials were not found. To set up Application Default Credentials, see https://cloud.google.com/docs/authentication/external/set-up-adc for more information.
litellm  | Traceback (most recent call last):
litellm  |   File "/usr/lib/python3.13/site-packages/litellm/main.py", line 557, in acompletion
litellm  |     response = await init_response
litellm  |                ^^^^^^^^^^^^^^^^^^^
litellm  |   File "/usr/lib/python3.13/site-packages/litellm/llms/vertex_ai/gemini/vertex_and_google_ai_studio_gemini.py", line 1707, in async_streaming
litellm  |     _auth_header, vertex_project = await self._ensure_access_token_async(
litellm  |                                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
litellm  |     ...<3 lines>...
litellm  |     )
litellm  |     ^
litellm  |   File "/usr/lib/python3.13/site-packages/litellm/llms/vertex_ai/vertex_llm_base.py", line 592, in _ensure_access_token_async
litellm  |     raise e
litellm  |   File "/usr/lib/python3.13/site-packages/litellm/llms/vertex_ai/vertex_llm_base.py", line 587, in _ensure_access_token_async
litellm  |     return await asyncify(self.get_access_token)(
litellm  |            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
litellm  |     ...<2 lines>...
litellm  |     )
litellm  |     ^
litellm  |   File "/usr/lib/python3.13/site-packages/litellm/litellm_core_utils/asyncify.py", line 57, in wrapper
litellm  |     return await anyio.to_thread.run_sync(
litellm  |            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
litellm  |     ...<3 lines>...
litellm  |     )
litellm  |     ^
litellm  |   File "/usr/lib/python3.13/site-packages/anyio/to_thread.py", line 56, in run_sync
litellm  |     return await get_async_backend().run_sync_in_worker_thread(
litellm  |            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
litellm  |         func, args, abandon_on_cancel=abandon_on_cancel, limiter=limiter
litellm  |         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
litellm  |     )
litellm  |     ^
litellm  |   File "/usr/lib/python3.13/site-packages/anyio/_backends/_asyncio.py", line 2461, in run_sync_in_worker_thread
litellm  |     return await future
litellm  |            ^^^^^^^^^^^^
litellm  |   File "/usr/lib/python3.13/site-packages/anyio/_backends/_asyncio.py", line 962, in run
litellm  |     result = context.run(func, *args)
litellm  |   File "/usr/lib/python3.13/site-packages/litellm/llms/vertex_ai/vertex_llm_base.py", line 503, in get_access_token
litellm  |     raise e
litellm  |   File "/usr/lib/python3.13/site-packages/litellm/llms/vertex_ai/vertex_llm_base.py", line 496, in get_access_token
litellm  |     _credentials, credential_project_id = self.load_auth(
litellm  |                                           ~~~~~~~~~~~~~~^
litellm  |         credentials=credentials, project_id=project_id
litellm  |         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
litellm  |     )
litellm  |     ^
litellm  |   File "/usr/lib/python3.13/site-packages/litellm/llms/vertex_ai/vertex_llm_base.py", line 114, in load_auth
litellm  |     creds, creds_project_id = self._credentials_from_default_auth(
litellm  |                               ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^
litellm  |         scopes=["https://www.googleapis.com/auth/cloud-platform"]
litellm  |         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
litellm  |     )
litellm  |     ^
litellm  |   File "/usr/lib/python3.13/site-packages/litellm/llms/vertex_ai/vertex_llm_base.py", line 160, in _credentials_from_default_auth
litellm  |     return google_auth.default(scopes=scopes)
litellm  |            ~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^
litellm  |   File "/usr/lib/python3.13/site-packages/google/auth/_default.py", line 685, in default
litellm  |     raise exceptions.DefaultCredentialsError(_CLOUD_SDK_MISSING_CREDENTIALS)
litellm  | google.auth.exceptions.DefaultCredentialsError: Your default credentials were not found. To set up Application Default Credentials, see https://cloud.google.com/docs/authentication/external/set-up-adc for more information.

Are you a ML Ops Team?

No

What LiteLLM version are you on ?

v1.77.3 OAS 3.1

Twitter / LinkedIn details

No response

[seagrayc]

If you prefix the model with gemini/ it will use gemini instead of vertex

model="gemini/gemini-2.5-flash-lite"

[jptrepo]

I'm getting the same as the OP.

My config.yaml is

• model_name: gemini-2.5-flash
  litellm_params:
  model: gemini/gemini-2.5-flash
  api_base: https://generativelanguage.googleapis.com/
  api_key: os.environ/GEMINI_API_KEY
• model_name: gemini-2.5-pro
  litellm_params:
  model: gemini/gemini-2.5-pro
  api_base: https://generativelanguage.googleapis.com/
  api_key: os.environ/GEMINI_API_KEY

Doing an API call gets me:
curl -X 'GET'
'DOMAIN/model/info'
-H 'accept: application/json'
-H 'x-litellm-api-key: KEY'

{
"model_name": "gemini-2.5-pro",
"litellm_params": {
"api_base": "https://generativelanguage.googleapis.com/",
"use_in_pass_through": false,
"use_litellm_proxy": false,
"merge_reasoning_content_in_choices": false,
"model": "gemini/gemini-2.5-pro"
},
"model_info": {
"id": "d62715ecac57c53e0ac93c33ccd5b78f2dd1a5a2ab274dfd169bea89322ec5bf",
"db_model": false,
"key": "gemini/gemini-2.5-pro",
"max_tokens": 65535,
"max_input_tokens": 1048576,
"max_output_tokens": 65535,
"input_cost_per_token": 0.00000125,
"cache_creation_input_token_cost": null,
"cache_read_input_token_cost": null,
"input_cost_per_character": null,
"input_cost_per_token_above_128k_tokens": null,
"input_cost_per_token_above_200k_tokens": 0.0000025,
"input_cost_per_query": null,
"input_cost_per_second": null,
"input_cost_per_audio_token": null,
"input_cost_per_token_batches": null,
"output_cost_per_token_batches": null,
"output_cost_per_token": 0.00001,
"output_cost_per_audio_token": null,
"output_cost_per_character": null,
"output_cost_per_reasoning_token": null,
"output_cost_per_token_above_128k_tokens": null,
"output_cost_per_character_above_128k_tokens": null,
"output_cost_per_token_above_200k_tokens": 0.000015,
"output_cost_per_second": null,
"output_cost_per_image": null,
"output_vector_size": null,
"citation_cost_per_token": null,
"litellm_provider": "gemini",
"mode": "chat",
"supports_system_messages": true,
"supports_response_schema": true,
"supports_vision": true,
"supports_function_calling": true,
"supports_tool_choice": true,
"supports_assistant_prefill": null,
"supports_prompt_caching": null,
"supports_audio_input": true,
"supports_audio_output": null,
"supports_pdf_input": true,
"supports_embedding_image_input": null,
"supports_native_streaming": null,
"supports_web_search": true,
"supports_url_context": null,
"supports_reasoning": true,
"supports_computer_use": null,
"search_context_cost_per_query": null,
"tpm": 800000,
"rpm": 2000,
"supported_openai_params": [
"temperature",
"top_p",
"max_tokens",
"max_completion_tokens",
"stream",
"tools",
"tool_choice",
"functions",
"response_format",
"n",
"stop",
"logprobs",
"frequency_penalty",
"modalities",
"parallel_tool_calls",
"web_search_options",
"reasoning_effort",
"thinking"
]
}
},

Yes I continue to get Vertex related errors.

[rawkintrevo]

Same error as op, model gemini/gemini-2.5-pro

After some debugging, I found, that I wasn't prepending gemini, just calling the model eg model='gemini-2.5-pro.

[rajesh-kanade]

I am facing similar issues .. just checking if this has been fixed recently

[helitra]

facing similar issues, no fix so far?

[guilyx]

Still happening!

[guilyx]

FYI @sanctimon, adding the gemini/ prefix to the model name already works, it's just counter intuitive IMO so i opened #19394

[ephb]

As far as I can tell the workaround does not work anymore. Old models that I added like this still work but the I can not add new models the same way without this issue showing up.

[github-actions]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs.