Skip to main content

Ctrl+K

Installation

Install SGLang

Backend Tutorial

DeepSeek Usage
Llama4 Usage
Sending Requests
OpenAI APIs - Completions
OpenAI APIs - Vision
OpenAI APIs - Embedding
SGLang Native APIs
Offline Engine API

Advanced Backend Configurations

Server Arguments
Sampling Parameters
Hyperparameter Tuning
Attention Backend

Supported Models

Large Language Models
Multimodal Language Models
Embedding Models
Reward Models
How to Support New Models

Advanced Features

Speculative Decoding
Structured Outputs
Tool and Function Calling
Reasoning Parser
Structured Outputs For Reasoning Models
Custom Chat Template
Quantization
LoRA Serving
PD Disaggregation

Frontend Tutorial

SGLang Frontend Language
Choices Methods in SGLang

SGLang Router

Router for Data Parallelism

References

General Guidance
Hardware Supports
Multi-Node Deployment
Performance Analysis & Optimization
- Benchmark and Profiling
- Measuring Model Accuracy in SGLang

Repository
Show source
Suggest edit
Open issue

.rst

Performance Analysis & Optimization

Performance Analysis & Optimization#

Benchmark and Profiling
Measuring Model Accuracy in SGLang

previous

Deploy On Kubernetes

next

Benchmark and Profiling

By SGLang Team

© Copyright 2023-2025, SGLang.

Last updated on May 26, 2025.