mirror of
https://github.com/BerriAI/litellm.git
synced 2025-04-26 11:14:04 +00:00
ddf77656fb
The problem we were having is non-admin users trying to use
`/engines/{model}/chat/completions` were getting an HTTP 401 error.
```shell
$ curl -sSL 'http://localhost:4000/engines/gpt-35-turbo-0125/chat/completions' \
--header "Authorization: Bearer ${LITELLM_KEY}" \
--header 'Content-Type: application/json' \
--data ' {
"model": "gpt-35-turbo-0125",
"messages": [
{
"role": "user",
"content": "Write a poem about LiteLLM"
}
]
}' \
| jq '.'
{
"error": {
"message": "Authentication Error, Only proxy admin can be used to generate, delete, update info for new keys/users/teams. Route=/engines/gpt-35-turbo-0125/chat/completions. Your role=unknown. Your user_id=someone@company.com",
"type": "auth_error",
"param": "None",
"code": 401
}
}
```
This seems to be related to code in `user_api_key_auth` that checks that the URL
matches a list of routes that are allowed for non-admin users, where the list of
routes is in `LiteLLMRoutes.openai_routes.value`. The problem is that the route
`/engines/{model}/chat/completions` is not in that list and furthermore, that
wouldn't even work if it were, because the comparison is done with
`request.url.path` and that will have the actual model name in it (e.g.:
`gpt-35-turbo-0125`), rather than `{model}`.
I added a new list `LiteLLMRoutes.openai_route_names` and added the route
**names** to that list. Then I added a check in `user_api_key_auth` to see if
the route name is in the list of route names.
|
||
|---|---|---|
| .. | ||
| _experimental | ||
| auth | ||
| db | ||
| example_config_yaml | ||
| hooks | ||
| proxy_load_test | ||
| queue | ||
| secret_managers | ||
| tests | ||
| .gitignore | ||
| __init__.py | ||
| _new_secret_config.yaml | ||
| _super_secret_config.yaml | ||
| _types.py | ||
| admin_ui.py | ||
| cached_logo.jpg | ||
| custom_callbacks.py | ||
| enterprise | ||
| health_check.py | ||
| lambda.py | ||
| llamaguard_prompt.txt | ||
| logo.jpg | ||
| openapi.json | ||
| otel_config.yaml | ||
| post_call_rules.py | ||
| proxy_cli.py | ||
| proxy_config.yaml | ||
| proxy_server.py | ||
| README.md | ||
| schema.prisma | ||
| start.sh | ||
| utils.py | ||
litellm-proxy
A local, fast, and lightweight OpenAI-compatible server to call 100+ LLM APIs.
usage
$ pip install litellm
$ litellm --model ollama/codellama
#INFO: Ollama running on http://0.0.0.0:8000
replace openai base
import openai # openai v1.0.0+
client = openai.OpenAI(api_key="anything",base_url="http://0.0.0.0:8000") # set proxy to base_url
# request sent to model set on litellm proxy, `litellm --model`
response = client.chat.completions.create(model="gpt-3.5-turbo", messages = [
{
"role": "user",
"content": "this is a test request, write a short poem"
}
])
print(response)
See how to call Huggingface,Bedrock,TogetherAI,Anthropic, etc.