Merge branch 'main' into litellm_ui_invite_links

litellm/integrations/email_templates/templates.py

@@ -0,0 +1,62 @@
+"""
+Email Templates used by the LiteLLM Email Service in slack_alerting.py
+"""
+
+KEY_CREATED_EMAIL_TEMPLATE = """
+                    <img src="{email_logo_url}" alt="LiteLLM Logo" width="150" height="50" />
+
+                    <p> Hi {recipient_email}, <br/>
+        
+                    I'm happy to provide you with an OpenAI Proxy API Key, loaded with ${key_budget} per month. <br /> <br />
+
+                    <b>
+                    Key: <pre>{key_token}</pre> <br>
+                    </b>
+
+                    <h2>Usage Example</h2>
+
+                    Detailed Documentation on <a href="https://docs.litellm.ai/docs/proxy/user_keys">Usage with OpenAI Python SDK, Langchain, LlamaIndex, Curl</a>
+
+                    <pre>
+
+                    import openai
+                    client = openai.OpenAI(
+                        api_key="{key_token}",
+                        base_url={{base_url}}
+                    )
+
+                    response = client.chat.completions.create(
+                        model="gpt-3.5-turbo", # model to send to the proxy
+                        messages = [
+                            {{
+                                "role": "user",
+                                "content": "this is a test request, write a short poem"
+                            }}
+                        ]
+                    )
+
+                    </pre>
+
+
+                    If you have any questions, please send an email to {email_support_contact} <br /> <br />
+
+                    Best, <br />
+                    The LiteLLM team <br />
+"""
+
+
+USER_INVITED_EMAIL_TEMPLATE = """
+                    <img src="{email_logo_url}" alt="LiteLLM Logo" width="150" height="50" />
+
+                    <p> Hi {recipient_email}, <br/>
+
+                    You were invited to use OpenAI Proxy API for team {team_name}  <br /> <br />
+
+                    <a href="{base_url}" style="display: inline-block; padding: 10px 20px; background-color: #87ceeb; color: #fff; text-decoration: none; border-radius: 20px;">Get Started here</a> <br /> <br />
+
+                    
+                    If you have any questions, please send an email to {email_support_contact} <br /> <br />
+
+                    Best, <br />
+                    The LiteLLM team <br />
+"""

litellm/integrations/slack_alerting.py

@@ -18,6 +18,7 @@ from litellm.proxy._types import WebhookEvent
 import random
 from typing import TypedDict
 from openai import APIError
+from .email_templates.templates import *
 
 import litellm.types
 from litellm.types.router import LiteLLM_Params
@@ -1190,8 +1191,12 @@ Model Info:
                 raise ValueError(
                     f"Trying to Customize Email Alerting\n {CommonProxyErrors.not_premium_user.value}"
                 )
+        return
 
-    async def send_key_created_email(self, webhook_event: WebhookEvent) -> bool:
+    async def send_key_created_or_user_invited_email(
+        self, webhook_event: WebhookEvent
+    ) -> bool:
+        try:
             from litellm.proxy.utils import send_email
 
             if self.alerting is None or "email" not in self.alerting:
@@ -1227,53 +1232,46 @@ Model Info:
             key_name = webhook_event.key_alias
             key_token = webhook_event.token
             key_budget = webhook_event.max_budget
+            base_url = os.getenv("PROXY_BASE_URL", "http://0.0.0.0:4000")
 
             email_html_content = "Alert from LiteLLM Server"
             if recipient_email is None:
                 verbose_proxy_logger.error(
-                "Trying to send email alert to no recipient", extra=webhook_event.dict()
-            )
-        email_html_content = f"""
-            <img src="{email_logo_url}" alt="LiteLLM Logo" width="150" height="50" />
-
-            <p> Hi {recipient_email}, <br/>
- 
-            I'm happy to provide you with an OpenAI Proxy API Key, loaded with ${key_budget} per month. <br /> <br />
-
-            <b>
-            Key: <pre>{key_token}</pre> <br>
-            </b>
-
-            <h2>Usage Example</h2>
-
-            Detailed Documentation on <a href="https://docs.litellm.ai/docs/proxy/user_keys">Usage with OpenAI Python SDK, Langchain, LlamaIndex, Curl</a>
-
-            <pre>
-
-            import openai
-            client = openai.OpenAI(
-                api_key="{key_token}",
-                base_url={os.getenv("PROXY_BASE_URL", "http://0.0.0.0:4000")}
+                    "Trying to send email alert to no recipient",
+                    extra=webhook_event.dict(),
                 )
 
-            response = client.chat.completions.create(
-                model="gpt-3.5-turbo", # model to send to the proxy
-                messages = [
-                    {{
-                        "role": "user",
-                        "content": "this is a test request, write a short poem"
-                    }}
-                ]
+            if webhook_event.event == "key_created":
+                email_html_content = KEY_CREATED_EMAIL_TEMPLATE.format(
+                    email_logo_url=email_logo_url,
+                    recipient_email=recipient_email,
+                    key_budget=key_budget,
+                    key_token=key_token,
+                    base_url=base_url,
+                    email_support_contact=email_support_contact,
+                )
+            elif webhook_event.event == "internal_user_created":
+                # GET TEAM NAME
+                team_id = webhook_event.team_id
+                team_name = "Default Team"
+                if team_id is not None and prisma_client is not None:
+                    team_row = await prisma_client.db.litellm_teamtable.find_unique(
+                        where={"team_id": team_id}
+                    )
+                    if team_row is not None:
+                        team_name = team_row.team_alias or "-"
+                email_html_content = USER_INVITED_EMAIL_TEMPLATE.format(
+                    email_logo_url=email_logo_url,
+                    recipient_email=recipient_email,
+                    team_name=team_name,
+                    base_url=base_url,
+                    email_support_contact=email_support_contact,
+                )
+            else:
+                verbose_proxy_logger.error(
+                    "Trying to send email alert on unknown webhook event",
+                    extra=webhook_event.model_dump(),
                 )
-
-            </pre>
-
-
-            If you have any questions, please send an email to {email_support_contact} <br /> <br />
-
-            Best, <br />
-            The LiteLLM team <br />
-            """
 
             payload = webhook_event.model_dump_json()
             email_event = {
@@ -1288,6 +1286,10 @@ Model Info:
                 html=email_event["html"],
             )
 
+            return True
+
+        except Exception as e:
+            verbose_proxy_logger.error("Error sending email alert %s", str(e))
             return False
 
     async def send_email_alert_using_smtp(self, webhook_event: WebhookEvent) -> bool:

litellm/model_prices_and_context_window_backup.json

@@ -380,6 +380,18 @@
         "output_cost_per_second": 0.0001, 
         "litellm_provider": "azure"
     },
+    "azure/gpt-4o": {
+        "max_tokens": 4096,
+        "max_input_tokens": 128000,
+        "max_output_tokens": 4096,
+        "input_cost_per_token": 0.000005,
+        "output_cost_per_token": 0.000015,
+        "litellm_provider": "azure",
+        "mode": "chat",
+        "supports_function_calling": true,
+        "supports_parallel_function_calling": true,
+        "supports_vision": true
+    },
     "azure/gpt-4-turbo-2024-04-09": {
         "max_tokens": 4096,
         "max_input_tokens": 128000,

litellm/proxy/_types.py

@@ -407,7 +407,9 @@ class ProxyChatCompletionRequest(LiteLLMBase):
     deployment_id: Optional[str] = None
     request_timeout: Optional[int] = None
 
-    model_config = ConfigDict(extra="allow")  # allow params not defined here, these fall in litellm.completion(**kwargs)
+    model_config = ConfigDict(
+        extra="allow"
+    )  # allow params not defined here, these fall in litellm.completion(**kwargs)
 
 
 class ModelInfoDelete(LiteLLMBase):
@@ -508,6 +510,7 @@ class GenerateKeyRequest(GenerateRequestBase):
     )  # {"gpt-4": 5.0, "gpt-3.5-turbo": 5.0}, defaults to {}
 
     model_config = ConfigDict(protected_namespaces=())
+    send_invite_email: Optional[bool] = None
 
 
 class GenerateKeyResponse(GenerateKeyRequest):
@@ -579,10 +582,24 @@ class NewUserRequest(GenerateKeyRequest):
     auto_create_key: bool = (
         True  # flag used for returning a key as part of the /user/new response
     )
+    send_invite_email: Optional[bool] = None
 
 
 class NewUserResponse(GenerateKeyResponse):
     max_budget: Optional[float] = None
+    user_email: Optional[str] = None
+    user_role: Optional[
+        Literal[
+            LitellmUserRoles.PROXY_ADMIN,
+            LitellmUserRoles.PROXY_ADMIN_VIEW_ONLY,
+            LitellmUserRoles.INTERNAL_USER,
+            LitellmUserRoles.INTERNAL_USER_VIEW_ONLY,
+            LitellmUserRoles.TEAM,
+            LitellmUserRoles.CUSTOMER,
+        ]
+    ] = None
+    teams: Optional[list] = None
+    organization_id: Optional[str] = None
 
 
 class UpdateUserRequest(GenerateRequestBase):
@@ -918,7 +935,9 @@ class KeyManagementSettings(LiteLLMBase):
 class TeamDefaultSettings(LiteLLMBase):
     team_id: str
 
-    model_config = ConfigDict(extra="allow")  # allow params not defined here, these fall in litellm.completion(**kwargs)
+    model_config = ConfigDict(
+        extra="allow"
+    )  # allow params not defined here, these fall in litellm.completion(**kwargs)
 
 
 class DynamoDBArgs(LiteLLMBase):
@@ -1112,6 +1131,7 @@ class LiteLLM_VerificationToken(LiteLLMBase):
 
     model_config = ConfigDict(protected_namespaces=())
 
+
 class LiteLLM_VerificationTokenView(LiteLLM_VerificationToken):
     """
     Combined view of litellm verification token + litellm team table (select values)
@@ -1285,6 +1305,7 @@ class WebhookEvent(CallInfo):
         "threshold_crossed",
         "projected_limit_exceeded",
         "key_created",
+        "internal_user_created",
         "spend_tracked",
     ]
     event_group: Literal["internal_user", "key", "team", "proxy", "customer"]

litellm/proxy/proxy_config.yaml

@@ -21,7 +21,7 @@ model_list:
 
 general_settings:
   master_key: sk-1234
-  alerting: ["slack"]
+  alerting: ["slack", "email"]
 
 litellm_settings:
   callbacks: custom_callbacks1.proxy_handler_instance

litellm/proxy/proxy_server.py

@@ -3217,6 +3217,9 @@ def _duration_in_seconds(duration: str):
 
 
 async def generate_key_helper_fn(
+    request_type: Literal[
+        "user", "key"
+    ],  # identifies if this request is from /user/new or /key/generate
     duration: Optional[str],
     models: list,
     aliases: dict,
@@ -3248,6 +3251,7 @@ async def generate_key_helper_fn(
     teams: Optional[list] = None,
     organization_id: Optional[str] = None,
     table_name: Optional[Literal["key", "user"]] = None,
+    send_invite_email: Optional[bool] = None,
 ):
     global prisma_client, custom_db_client, user_api_key_cache, litellm_proxy_admin_name, premium_user
 
@@ -3282,7 +3286,7 @@ async def generate_key_helper_fn(
     permissions_json = json.dumps(permissions)
     metadata_json = json.dumps(metadata)
     model_max_budget_json = json.dumps(model_max_budget)
-    user_role = user_role or "app_user"
+    user_role = user_role
     tpm_limit = tpm_limit
     rpm_limit = rpm_limit
     allowed_cache_controls = allowed_cache_controls
@@ -3348,7 +3352,7 @@ async def generate_key_helper_fn(
                 "get_spend_routes" in saved_token["permissions"]
                 and premium_user != True
             ):
-                raise Exception(
+                raise ValueError(
                     "get_spend_routes permission is only available for LiteLLM Enterprise users"
                 )
 
@@ -3405,6 +3409,10 @@ async def generate_key_helper_fn(
 
     # Add budget related info in key_data - this ensures it's returned
     key_data["budget_id"] = budget_id
+
+    if request_type == "user":
+        # if this is a /user/new request update the key_date with user_data fields
+        key_data.update(user_data)
     return key_data
 
 
@@ -3752,6 +3760,7 @@ async def startup_event():
             )
         asyncio.create_task(
             generate_key_helper_fn(
+                request_type="user",
                 duration=None,
                 models=[],
                 aliases={},
@@ -3774,6 +3783,7 @@ async def startup_event():
         # add proxy budget to db in the user table
         asyncio.create_task(
             generate_key_helper_fn(
+                request_type="user",
                 user_id=litellm_proxy_budget_name,
                 duration=None,
                 models=[],
@@ -3796,7 +3806,13 @@ async def startup_event():
     if custom_db_client is not None and master_key is not None:
         # add master key to db
         await generate_key_helper_fn(
-            duration=None, models=[], aliases={}, config={}, spend=0, token=master_key
+            request_type="key",
+            duration=None,
+            models=[],
+            aliases={},
+            config={},
+            spend=0,
+            token=master_key,
         )
 
     ### CHECK IF VIEW EXISTS ###
@@ -6004,6 +6020,7 @@ async def generate_key_fn(
     - aliases: Optional[dict] - Any alias mappings, on top of anything in the config.yaml model list. - https://docs.litellm.ai/docs/proxy/virtual_keys#managing-auth---upgradedowngrade-models
     - config: Optional[dict] - any key-specific configs, overrides config in config.yaml
     - spend: Optional[int] - Amount spent by key. Default is 0. Will be updated by proxy whenever key is used. https://docs.litellm.ai/docs/proxy/virtual_keys#managing-auth---tracking-spend
+    - send_invite_email: Optional[bool] - Whether to send an invite email to the user_id, with the generate key
     - max_budget: Optional[float] - Specify max budget for a given key.
     - max_parallel_requests: Optional[int] - Rate limit a user based on the number of parallel requests. Raises 429 error, if user's parallel requests > x.
     - metadata: Optional[dict] - Metadata for key, store information for key. Example metadata = {"team": "core-infra", "app": "app2", "email": "ishaan@berri.ai" }
@@ -6132,11 +6149,19 @@ async def generate_key_fn(
         if "budget_duration" in data_json:
             data_json["key_budget_duration"] = data_json.pop("budget_duration", None)
 
-        response = await generate_key_helper_fn(**data_json, table_name="key")
+        response = await generate_key_helper_fn(
+            request_type="key", **data_json, table_name="key"
+        )
 
         response["soft_budget"] = (
             data.soft_budget
         )  # include the user-input soft budget in the response
+
+        if data.send_invite_email is True:
+            if "email" not in general_settings.get("alerting", []):
+                raise ValueError(
+                    "Email alerting not setup on config.yaml. Please set `alerting=['email']. \nDocs: https://docs.litellm.ai/docs/proxy/email`"
+                )
             event = WebhookEvent(
                 event="key_created",
                 event_group="key",
@@ -6151,7 +6176,7 @@ async def generate_key_fn(
 
             # If user configured email alerting - send an Email letting their end-user know the key was created
             asyncio.create_task(
-            proxy_logging_obj.slack_alerting_instance.send_key_created_email(
+                proxy_logging_obj.slack_alerting_instance.send_key_created_or_user_invited_email(
                     webhook_event=event,
                 )
             )
@@ -8116,6 +8141,7 @@ async def new_user(data: NewUserRequest):
     - teams: Optional[list] - specify a list of team id's a user belongs to.
     - organization_id: Optional[str] - specify the org a user belongs to.
     - user_email: Optional[str] - Specify a user email.
+    - send_invite_email: Optional[bool] - Specify if an invite email should be sent.
     - user_role: Optional[str] - Specify a user role - "admin", "app_owner", "app_user"
     - max_budget: Optional[float] - Specify max budget for a given user.
     - models: Optional[list] - Model_name's a user is allowed to call. (if empty, key is allowed to call all models)
@@ -8137,7 +8163,7 @@ async def new_user(data: NewUserRequest):
         data_json["table_name"] = (
             "user"  # only create a user, don't create key if 'auto_create_key' set to False
         )
-    response = await generate_key_helper_fn(**data_json)
+    response = await generate_key_helper_fn(request_type="user", **data_json)
 
     # Admin UI Logic
     # if team_id passed add this user to the team
@@ -8152,11 +8178,42 @@ async def new_user(data: NewUserRequest):
                 ),
             )
         )
+
+    if data.send_invite_email is True:
+        # check if user has setup email alerting
+        if "email" not in general_settings.get("alerting", []):
+            raise ValueError(
+                "Email alerting not setup on config.yaml. Please set `alerting=['email']. \nDocs: https://docs.litellm.ai/docs/proxy/email`"
+            )
+
+        event = WebhookEvent(
+            event="internal_user_created",
+            event_group="internal_user",
+            event_message=f"Welcome to LiteLLM Proxy",
+            token=response.get("token", ""),
+            spend=response.get("spend", 0.0),
+            max_budget=response.get("max_budget", 0.0),
+            user_id=response.get("user_id", None),
+            user_email=response.get("user_email", None),
+            team_id=response.get("team_id", "Default Team"),
+            key_alias=response.get("key_alias", None),
+        )
+
+        # If user configured email alerting - send an Email letting their end-user know the key was created
+        asyncio.create_task(
+            proxy_logging_obj.slack_alerting_instance.send_key_created_or_user_invited_email(
+                webhook_event=event,
+            )
+        )
+
     return NewUserResponse(
         key=response.get("token", ""),
         expires=response.get("expires", None),
         max_budget=response["max_budget"],
         user_id=response["user_id"],
+        user_role=response.get("user_role", None),
+        user_email=response.get("user_email", None),
+        teams=response.get("teams", None),
         team_id=response.get("team_id", None),
         metadata=response.get("metadata", None),
         models=response.get("models", None),
@@ -8213,11 +8270,13 @@ async def user_auth(request: Request):
     if response is not None:
         user_id = response.user_id
         response = await generate_key_helper_fn(
-            **{"duration": "24hr", "models": [], "aliases": {}, "config": {}, "spend": 0, "user_id": user_id}  # type: ignore
+            request_type="key",
+            **{"duration": "24hr", "models": [], "aliases": {}, "config": {}, "spend": 0, "user_id": user_id},  # type: ignore
         )
     else:  ### else - create new user
         response = await generate_key_helper_fn(
-            **{"duration": "24hr", "models": [], "aliases": {}, "config": {}, "spend": 0, "user_email": user_email}  # type: ignore
+            request_type="key",
+            **{"duration": "24hr", "models": [], "aliases": {}, "config": {}, "spend": 0, "user_email": user_email},  # type: ignore
         )
 
     base_url = os.getenv("LITELLM_HOSTED_UI", "https://dashboard.litellm.ai/")
@@ -11709,7 +11768,8 @@ async def login(request: Request):
         )
         if os.getenv("DATABASE_URL") is not None:
             response = await generate_key_helper_fn(
-                **{"user_role": LitellmUserRoles.PROXY_ADMIN, "duration": "2hr", "key_max_budget": 5, "models": [], "aliases": {}, "config": {}, "spend": 0, "user_id": key_user_id, "team_id": "litellm-dashboard"}  # type: ignore
+                request_type="key",
+                **{"user_role": LitellmUserRoles.PROXY_ADMIN, "duration": "2hr", "key_max_budget": 5, "models": [], "aliases": {}, "config": {}, "spend": 0, "user_id": key_user_id, "team_id": "litellm-dashboard"},  # type: ignore
             )
         else:
             raise ProxyException(
@@ -11810,6 +11870,7 @@ async def onboarding(invite_link: str):
     response = await generate_key_helper_fn(
         **{
             "user_role": LitellmUserRoles.PROXY_ADMIN,
+            request_type="key",
             "duration": "2hr",
             "key_max_budget": 5,
             "models": [],
@@ -12125,8 +12186,11 @@ async def auth_callback(request: Request):
     verbose_proxy_logger.info(
         f"user_defined_values for creating ui key: {user_defined_values}"
     )
+
+    default_ui_key_values.update(user_defined_values)
+    default_ui_key_values["request_type"] = "key"
     response = await generate_key_helper_fn(
-        **default_ui_key_values, **user_defined_values  # type: ignore
+        **default_ui_key_values,  # type: ignore
     )
     key = response["token"]  # type: ignore
     user_id = response["user_id"]  # type: ignore
@@ -13231,7 +13295,7 @@ async def health_services_endpoint(
 
             # use create task - this can take 10 seconds. don't keep ui users waiting for notification to check their email
             asyncio.create_task(
-                proxy_logging_obj.slack_alerting_instance.send_key_created_email(
+                proxy_logging_obj.slack_alerting_instance.send_key_created_or_user_invited_email(
                     webhook_event=webhook_event
                 )
             )

litellm/tests/test_key_generate_prisma.py

@@ -1912,7 +1912,8 @@ async def test_key_with_no_permissions(prisma_client):
     await litellm.proxy.proxy_server.prisma_client.connect()
     try:
         response = await generate_key_helper_fn(
-            **{"duration": "1hr", "key_max_budget": 0, "models": [], "aliases": {}, "config": {}, "spend": 0, "user_id": "ishaan", "team_id": "litellm-dashboard"}  # type: ignore
+            request_type="key",
+            **{"duration": "1hr", "key_max_budget": 0, "models": [], "aliases": {}, "config": {}, "spend": 0, "user_id": "ishaan", "team_id": "litellm-dashboard"},  # type: ignore
         )
 
         print(response)

model_prices_and_context_window.json

@@ -380,6 +380,18 @@
         "output_cost_per_second": 0.0001, 
         "litellm_provider": "azure"
     },
+    "azure/gpt-4o": {
+        "max_tokens": 4096,
+        "max_input_tokens": 128000,
+        "max_output_tokens": 4096,
+        "input_cost_per_token": 0.000005,
+        "output_cost_per_token": 0.000015,
+        "litellm_provider": "azure",
+        "mode": "chat",
+        "supports_function_calling": true,
+        "supports_parallel_function_calling": true,
+        "supports_vision": true
+    },
     "azure/gpt-4-turbo-2024-04-09": {
         "max_tokens": 4096,
         "max_input_tokens": 128000,