Benchmarks

Benchmarks for LiteLLM Gateway (Proxy Server) tested against a fake OpenAI endpoint.

Use this config for testing:

Note: we're currently migrating to aiohttp which has 10x higher throughput. We recommend using the aiohttp_openai/ provider for load testing.

model_list:
  - model_name: "fake-openai-endpoint"
    litellm_params:
      model: aiohttp_openai/any
      api_base: https://your-fake-openai-endpoint.com/chat/completions
      api_key: "test"

1 Instance LiteLLM Proxy

Metric	Litellm Proxy (1 Instance)
Median Latency (ms)	110
RPS	250

Key Findings

• Single instance: 250 RPS @ 100ms latency
• 4 LiteLLM instances: 1000 RPS @ 100ms latency

2 Instances

Adding 1 instance, will double the RPS and maintain the 100ms-110ms median latency.

Metric	Litellm Proxy (2 Instances)
Median Latency (ms)	100
RPS	500

Logging Callbacks

GCS Bucket Logging

Using GCS Bucket has no impact on latency, RPS compared to Basic Litellm Proxy

Metric	Basic Litellm Proxy	LiteLLM Proxy with GCS Bucket Logging
RPS	1133.2	1137.3
Median Latency (ms)	140	138

LangSmith logging

Using LangSmith has no impact on latency, RPS compared to Basic Litellm Proxy

Metric	Basic Litellm Proxy	LiteLLM Proxy with LangSmith
RPS	1133.2	1135
Median Latency (ms)	140	132

Locust Settings

• 2500 Users
• 100 user Ramp Up