mirror of
https://github.com/tiennm99/litellm.git
synced 2026-07-19 04:20:26 +00:00
Alexsander HamirandGitHub
eb5031da1e
* Cache realtime websocket request body Move the realtime request payload builder out of the websocket handler and wrap it with an LRU cache so repeated connections reuse the same bytes object. This keeps the JSON formatting cost down while bounding memory usage. * Optimize realtime websocket caching Refactored /v1/realtime to use cached helpers for both the JSON body and query params, introduced a reusable request-scope template, and optimized header handling to avoid redundant work. * Refine realtime websocket header handling * Reuse websocket scope headers in auth * Refactor realtime request body helper Move the realtime request body formatter into proxy common utils so it can be reused across modules. Reuse it in the websocket auth flow to share LRU caching and avoid ad hoc byte builders. * fix: revert to old pattern The old pattern was necessary, we can just return the optimized function instead. * Reuse SSL context for realtime Create a shared SSLContext for OpenAI realtime websocket dials and pass it into websockets.connect so we stop re-reading verify paths on every session. * feat: reuse shared TLS context for realtime websockets - add `SHARED_REALTIME_SSL_CONTEXT` helper so all realtime websocket clients share the same TLS settings - wire the shared context into OpenAI, Azure, custom HTTPX handlers, and realtime health checks - update realtime tests to assert that the expected SSL context is passed to `websockets.connect` This keeps TLS configuration consistent and avoids recreating SSL contexts per connection. * Reuse HTTP SSL context for realtime Remove the standalone realtime SSL helper, expose a shared context directly from the HTTP handler, and point all realtime websocket clients and tests to it. Add the websocket header comparison tool. * Lazy-load shared realtime SSL context Fix circular imports introduced by eagerly instantiating the shared TLS context. Make the HTTP handler lazily create the context and have realtime clients/tests fetch it on demand, keeping configuration consistent without breaking startup. * add: unit test for realtime LRU caches * fix: merge conflict with imports
🚅 LiteLLM
Call all LLM APIs using the OpenAI format [Bedrock, Huggingface, VertexAI, TogetherAI, Azure, OpenAI, Groq etc.]
LiteLLM Proxy Server (LLM Gateway) | Hosted Proxy (Preview) | Enterprise Tier
LiteLLM manages:
- Translate inputs to provider's
completion,embedding, andimage_generationendpoints - Consistent output, text responses will always be available at
['choices'][0]['message']['content'] - Retry/fallback logic across multiple deployments (e.g. Azure/OpenAI) - Router
- Set Budgets & Rate limits per project, api key, model LiteLLM Proxy Server (LLM Gateway)
LiteLLM Performance: 8ms P95 latency at 1k RPS (See benchmarks here)
Jump to LiteLLM Proxy (LLM Gateway) Docs
Jump to Supported LLM Providers
🚨 Stable Release: Use docker images with the -stable tag. These have undergone 12 hour load tests, before being published. More information about the release cycle here
Support for more providers. Missing a provider or LLM Platform, raise a feature request.
Usage (Docs)
Important
LiteLLM v1.0.0 now requires
openai>=1.0.0. Migration guide here LiteLLM v1.40.14+ now requirespydantic>=2.0.0. No changes required.
pip install litellm
from litellm import completion
import os
## set ENV variables
os.environ["OPENAI_API_KEY"] = "your-openai-key"
os.environ["ANTHROPIC_API_KEY"] = "your-anthropic-key"
messages = [{ "content": "Hello, how are you?","role": "user"}]
# openai call
response = completion(model="openai/gpt-4o", messages=messages)
# anthropic call
response = completion(model="anthropic/claude-sonnet-4-20250514", messages=messages)
print(response)
Response (OpenAI Format)
{
"id": "chatcmpl-1214900a-6cdd-4148-b663-b5e2f642b4de",
"created": 1751494488,
"model": "claude-sonnet-4-20250514",
"object": "chat.completion",
"system_fingerprint": null,
"choices": [
{
"finish_reason": "stop",
"index": 0,
"message": {
"content": "Hello! I'm doing well, thank you for asking. I'm here and ready to help with whatever you'd like to discuss or work on. How are you doing today?",
"role": "assistant",
"tool_calls": null,
"function_call": null
}
}
],
"usage": {
"completion_tokens": 39,
"prompt_tokens": 13,
"total_tokens": 52,
"completion_tokens_details": null,
"prompt_tokens_details": {
"audio_tokens": null,
"cached_tokens": 0
},
"cache_creation_input_tokens": 0,
"cache_read_input_tokens": 0
}
}
Call any model supported by a provider, with model=<provider_name>/<model_name>. There might be provider-specific details here, so refer to provider docs for more information
Async (Docs)
from litellm import acompletion
import asyncio
async def test_get_response():
user_message = "Hello, how are you?"
messages = [{"content": user_message, "role": "user"}]
response = await acompletion(model="openai/gpt-4o", messages=messages)
return response
response = asyncio.run(test_get_response())
print(response)
Streaming (Docs)
LiteLLM supports streaming the model response back, pass stream=True to get a streaming iterator in response.
Streaming is supported for all models (Bedrock, Huggingface, TogetherAI, Azure, OpenAI, etc.)
from litellm import completion
messages = [{"content": "Hello, how are you?", "role": "user"}]
# gpt-4o
response = completion(model="openai/gpt-4o", messages=messages, stream=True)
for part in response:
print(part.choices[0].delta.content or "")
# claude sonnet 4
response = completion('anthropic/claude-sonnet-4-20250514', messages, stream=True)
for part in response:
print(part)
Response chunk (OpenAI Format)
{
"id": "chatcmpl-fe575c37-5004-4926-ae5e-bfbc31f356ca",
"created": 1751494808,
"model": "claude-sonnet-4-20250514",
"object": "chat.completion.chunk",
"system_fingerprint": null,
"choices": [
{
"finish_reason": null,
"index": 0,
"delta": {
"provider_specific_fields": null,
"content": "Hello",
"role": "assistant",
"function_call": null,
"tool_calls": null,
"audio": null
},
"logprobs": null
}
],
"provider_specific_fields": null,
"stream_options": null,
"citations": null
}
Logging Observability (Docs)
LiteLLM exposes pre defined callbacks to send data to Lunary, MLflow, Langfuse, DynamoDB, s3 Buckets, Helicone, Promptlayer, Traceloop, Athina, Slack
from litellm import completion
## set env variables for logging tools (when using MLflow, no API key set up is required)
os.environ["LUNARY_PUBLIC_KEY"] = "your-lunary-public-key"
os.environ["HELICONE_API_KEY"] = "your-helicone-auth-key"
os.environ["LANGFUSE_PUBLIC_KEY"] = ""
os.environ["LANGFUSE_SECRET_KEY"] = ""
os.environ["ATHINA_API_KEY"] = "your-athina-api-key"
os.environ["OPENAI_API_KEY"] = "your-openai-key"
# set callbacks
litellm.success_callback = ["lunary", "mlflow", "langfuse", "athina", "helicone"] # log input/output to lunary, langfuse, supabase, athina, helicone etc
#openai call
response = completion(model="openai/gpt-4o", messages=[{"role": "user", "content": "Hi 👋 - i'm openai"}])
LiteLLM Proxy Server (LLM Gateway) - (Docs)
Track spend + Load Balance across multiple projects
The proxy provides:
📖 Proxy Endpoints - Swagger Docs
Quick Start Proxy - CLI
pip install 'litellm[proxy]'
Step 1: Start litellm proxy
$ litellm --model huggingface/bigcode/starcoder
#INFO: Proxy running on http://0.0.0.0:4000
Step 2: Make ChatCompletions Request to Proxy
Important
import openai # openai v1.0.0+
client = openai.OpenAI(api_key="anything",base_url="http://0.0.0.0:4000") # 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)
Proxy Key Management (Docs)
Connect the proxy with a Postgres DB to create proxy keys
# Get the code
git clone https://github.com/BerriAI/litellm
# Go to folder
cd litellm
# Add the master key - you can change this after setup
echo 'LITELLM_MASTER_KEY="sk-1234"' > .env
# Add the litellm salt key - you cannot change this after adding a model
# It is used to encrypt / decrypt your LLM API Key credentials
# We recommend - https://1password.com/password-generator/
# password generator to get a random hash for litellm salt key
echo 'LITELLM_SALT_KEY="sk-1234"' >> .env
source .env
# Start
docker compose up
UI on /ui on your proxy server
Set budgets and rate limits across multiple projects
POST /key/generate
Request
curl 'http://0.0.0.0:4000/key/generate' \
--header 'Authorization: Bearer sk-1234' \
--header 'Content-Type: application/json' \
--data-raw '{"models": ["gpt-3.5-turbo", "gpt-4", "claude-2"], "duration": "20m","metadata": {"user": "ishaan@berri.ai", "team": "core-infra"}}'
Expected Response
{
"key": "sk-kdEXbIqZRwEeEiHwdg7sFA", # Bearer token
"expires": "2023-11-19T01:38:25.838000+00:00" # datetime object
}
Supported Providers (Website Supported Models | Docs)
Run in Developer mode
Services
- Setup .env file in root
- Run dependant services
docker-compose up db prometheus
Backend
- (In root) create virtual environment
python -m venv .venv - Activate virtual environment
source .venv/bin/activate - Install dependencies
pip install -e ".[all]" - Start proxy backend
python litellm/proxy_cli.py
Frontend
- Navigate to
ui/litellm-dashboard - Install dependencies
npm install - Run
npm run devto start the dashboard
Enterprise
For companies that need better security, user management and professional support
This covers:
- ✅ Features under the LiteLLM Commercial License:
- ✅ Feature Prioritization
- ✅ Custom Integrations
- ✅ Professional Support - Dedicated discord + slack
- ✅ Custom SLAs
- ✅ Secure access with Single Sign-On
Contributing
We welcome contributions to LiteLLM! Whether you're fixing bugs, adding features, or improving documentation, we appreciate your help.
Quick Start for Contributors
This requires poetry to be installed.
git clone https://github.com/BerriAI/litellm.git
cd litellm
make install-dev # Install development dependencies
make format # Format your code
make lint # Run all linting checks
make test-unit # Run unit tests
make format-check # Check formatting only
For detailed contributing guidelines, see CONTRIBUTING.md.
Code Quality / Linting
LiteLLM follows the Google Python Style Guide.
Our automated checks include:
- Black for code formatting
- Ruff for linting and code quality
- MyPy for type checking
- Circular import detection
- Import safety checks
All these checks must pass before your PR can be merged.
Support / talk with founders
- Schedule Demo 👋
- Community Discord 💭
- Community Slack 💭
- Our numbers 📞 +1 (770) 8783-106 / +1 (412) 618-6238
- Our emails ✉️ ishaan@berri.ai / krrish@berri.ai
Why did we build this
- Need for simplicity: Our code started to get extremely complicated managing & translating calls between Azure, OpenAI and Cohere.