# SGLang Documentation

## Docs

- [Unlimited-OCR](https://docs.sglang.io/cookbook/autoregressive/Baidu/Unlimited-OCR.md): Deploy Baidu Unlimited-OCR with SGLang for long document OCR using prefill-aware sliding-window attention.
- [Ornith-1.0](https://docs.sglang.io/cookbook/autoregressive/DeepReinforce/Ornith-1.0.md): Deploy DeepReinforce Ornith-1.0 with SGLang - a self-improving agentic-coding model family with 397B, 35B, and 9B checkpoints plus FP8 and GGUF variants.
- [DeepSeek-Math-V2](https://docs.sglang.io/cookbook/autoregressive/DeepSeek/DeepSeek-Math-V2.md)
- [DeepSeek-OCR](https://docs.sglang.io/cookbook/autoregressive/DeepSeek/DeepSeek-OCR.md)
- [DeepSeek-OCR-2](https://docs.sglang.io/cookbook/autoregressive/DeepSeek/DeepSeek-OCR-2.md)
- [DeepSeek-R1](https://docs.sglang.io/cookbook/autoregressive/DeepSeek/DeepSeek-R1.md)
- [DeepSeek-V3](https://docs.sglang.io/cookbook/autoregressive/DeepSeek/DeepSeek-V3.md)
- [DeepSeek-V3.1](https://docs.sglang.io/cookbook/autoregressive/DeepSeek/DeepSeek-V3_1.md)
- [DeepSeek-V3.2](https://docs.sglang.io/cookbook/autoregressive/DeepSeek/DeepSeek-V3_2.md)
- [DeepSeek-V4](https://docs.sglang.io/cookbook/autoregressive/DeepSeek/DeepSeek-V4.md): Deploy DeepSeek-V4 with SGLang — verified launch commands, benchmarks, and tuning for the Flash (284B) and Pro (1.6T) Mixture-of-Experts models.
- [Ernie4.5](https://docs.sglang.io/cookbook/autoregressive/Ernie/Ernie4.5.md)
- [Ernie4.5-VL](https://docs.sglang.io/cookbook/autoregressive/Ernie/Ernie4.5-VL.md)
- [Chroma-1.0](https://docs.sglang.io/cookbook/autoregressive/FlashLabs/Chroma1.0.md)
- [GLM-4.5](https://docs.sglang.io/cookbook/autoregressive/GLM/GLM-4.5.md)
- [GLM-4.5V](https://docs.sglang.io/cookbook/autoregressive/GLM/GLM-4.5V.md)
- [GLM-4.6](https://docs.sglang.io/cookbook/autoregressive/GLM/GLM-4.6.md)
- [GLM-4.6V](https://docs.sglang.io/cookbook/autoregressive/GLM/GLM-4.6V.md)
- [GLM-4.7](https://docs.sglang.io/cookbook/autoregressive/GLM/GLM-4.7.md)
- [GLM-4.7-Flash](https://docs.sglang.io/cookbook/autoregressive/GLM/GLM-4.7-Flash.md)
- [GLM-5](https://docs.sglang.io/cookbook/autoregressive/GLM/GLM-5.md)
- [GLM-5.1](https://docs.sglang.io/cookbook/autoregressive/GLM/GLM-5.1.md)
- [GLM-5.2](https://docs.sglang.io/cookbook/autoregressive/GLM/GLM-5.2.md): Deploy GLM-5.2 with SGLang — Z.ai's DeepSeek-Sparse-Attention (DSA) Mixture-of-Experts model with MTP speculative decoding and 1M context, on H200, B200, B300, GB300, and AMD MI300X/MI325X/MI355X.
- [GLM Glyph](https://docs.sglang.io/cookbook/autoregressive/GLM/GLM-Glyph.md)
- [GLM-OCR](https://docs.sglang.io/cookbook/autoregressive/GLM/GLM-OCR.md)
- [DiffusionGemma](https://docs.sglang.io/cookbook/autoregressive/Google/DiffusionGemma.md)
- [Gemma 4](https://docs.sglang.io/cookbook/autoregressive/Google/Gemma4.md)
- [LLaDA 2.1](https://docs.sglang.io/cookbook/autoregressive/InclusionAI/LLaDA-2.1.md)
- [Ling-2.5-1T](https://docs.sglang.io/cookbook/autoregressive/InclusionAI/Ling-2.5-1T.md)
- [Ling-2.6](https://docs.sglang.io/cookbook/autoregressive/InclusionAI/Ling-2.6.md)
- [Ring-2.5-1T](https://docs.sglang.io/cookbook/autoregressive/InclusionAI/Ring-2.5-1T.md)
- [Ring-2.6-1T](https://docs.sglang.io/cookbook/autoregressive/InclusionAI/Ring-2.6-1T.md)
- [Intern-S1](https://docs.sglang.io/cookbook/autoregressive/InternLM/Intern-S1.md)
- [Intern-S2-Preview](https://docs.sglang.io/cookbook/autoregressive/InternLM/Intern-S2-Preview.md)
- [InternVL3.5](https://docs.sglang.io/cookbook/autoregressive/InternVL/InternVL3.5.md)
- [Jina-reranker-m0](https://docs.sglang.io/cookbook/autoregressive/Jina/Jina-reranker-m0.md)
- [LFM2.5](https://docs.sglang.io/cookbook/autoregressive/LiquidAI/LFM2.5.md): Deploy Liquid AI's LFM2.5 with SGLang — hybrid gated short conv + GQA models from 350M to the 8B-A1B MoE, plus LFM2.5-VL vision, with reasoning and Pythonic tool calling.
- [Llama-3.1](https://docs.sglang.io/cookbook/autoregressive/Llama/Llama3.1.md)
- [Llama-3.3-70B](https://docs.sglang.io/cookbook/autoregressive/Llama/Llama3.3-70B.md)
- [Llama 4](https://docs.sglang.io/cookbook/autoregressive/Llama/Llama4.md)
- [LongCat-2.0](https://docs.sglang.io/cookbook/autoregressive/Meituan/LongCat-2.0.md): Deploy LongCat-2.0-FP8 with SGLang - config-driven recipes for Meituan's 1.6T sparse MoE model on B300, B200, H200, and H20 GPUs.
- [MiniMax-M2](https://docs.sglang.io/cookbook/autoregressive/MiniMax/MiniMax-M2.md)
- [MiniMax-M2.5](https://docs.sglang.io/cookbook/autoregressive/MiniMax/MiniMax-M2.5.md)
- [MiniMax-M2.7](https://docs.sglang.io/cookbook/autoregressive/MiniMax/MiniMax-M2.7.md)
- [MiniMax-M3](https://docs.sglang.io/cookbook/autoregressive/MiniMax/MiniMax-M3.md): Deploy MiniMax-M3 with SGLang — a ~428B-param (23B activated) multimodal Mixture-of-Experts reasoning model with MiniMax Sparse Attention and 1M context, MXFP8 on NVIDIA Blackwell & AMD Instinct, bf16 on Hopper.
- [Devstral 2 (Mistral)](https://docs.sglang.io/cookbook/autoregressive/Mistral/Devstral-2.md)
- [Ministral-3](https://docs.sglang.io/cookbook/autoregressive/Mistral/Ministral-3.md)
- [Mistral Medium 3.5](https://docs.sglang.io/cookbook/autoregressive/Mistral/Mistral-Medium-3.5.md)
- [Mistral Small 4](https://docs.sglang.io/cookbook/autoregressive/Mistral/Mistral-Small-4.md)
- [Kimi-K2](https://docs.sglang.io/cookbook/autoregressive/Moonshotai/Kimi-K2.md)
- [Kimi-K2.5](https://docs.sglang.io/cookbook/autoregressive/Moonshotai/Kimi-K2.5.md)
- [Kimi-K2.6](https://docs.sglang.io/cookbook/autoregressive/Moonshotai/Kimi-K2.6.md)
- [Kimi-K2.7-Code](https://docs.sglang.io/cookbook/autoregressive/Moonshotai/Kimi-K2.7-Code.md): Deploy Kimi-K2.7-Code with SGLang for coding-focused agentic workflows, thinking output, tool calling, and multimodal input.
- [Kimi-Linear](https://docs.sglang.io/cookbook/autoregressive/Moonshotai/Kimi-Linear.md)
- [Nemotron3-Nano](https://docs.sglang.io/cookbook/autoregressive/NVIDIA/Nemotron3-Nano.md)
- [Nemotron 3 Nano Omni](https://docs.sglang.io/cookbook/autoregressive/NVIDIA/Nemotron3-Nano-Omni.md)
- [NVIDIA Nemotron3-Super](https://docs.sglang.io/cookbook/autoregressive/NVIDIA/Nemotron3-Super.md)
- [NVIDIA Nemotron3-Ultra](https://docs.sglang.io/cookbook/autoregressive/NVIDIA/Nemotron3-Ultra.md): Deploy NVIDIA Nemotron3-Ultra with SGLang - 550B hybrid MoE model (55B active) with 1M context window, BF16/NVFP4 support, built for long-running autonomous agents.
- [GPT-OSS](https://docs.sglang.io/cookbook/autoregressive/OpenAI/GPT-OSS.md)
- [MiniCPM-V 4.6](https://docs.sglang.io/cookbook/autoregressive/OpenBMB/MiniCPM-V-4_6.md)
- [Laguna-M.1](https://docs.sglang.io/cookbook/autoregressive/Poolside/Laguna-M.1.md): Deploy poolside's Laguna-M.1 — a 225B-parameter Mixture-of-Experts model (23B active) for agentic coding — with SGLang on NVIDIA H200, B200, B300, GB200, and GB300, across BF16, FP8, and NVFP4.
- [Laguna-S-2.1](https://docs.sglang.io/cookbook/autoregressive/Poolside/Laguna-S-2.1.md): Deploy poolside's Laguna-S-2.1 — a 118B hybrid-SWA Mixture-of-Experts model (8B active) for agentic coding — with SGLang on NVIDIA H200, B300, and GB300 in BF16, FP8, NVFP4, and INT4.
- [Laguna-XS-2.1](https://docs.sglang.io/cookbook/autoregressive/Poolside/Laguna-XS-2.1.md): Deploy poolside's Laguna-XS-2.1 — a 33B hybrid-SWA Mixture-of-Experts model (3B active) for agentic coding — with SGLang on NVIDIA H200, B300, and GB300 in BF16, FP8, NVFP4, and INT4.
- [Laguna-XS.2](https://docs.sglang.io/cookbook/autoregressive/Poolside/Laguna-XS.2.md)
- [Qwen2.5-VL](https://docs.sglang.io/cookbook/autoregressive/Qwen/Qwen2.5-VL.md)
- [Qwen3](https://docs.sglang.io/cookbook/autoregressive/Qwen/Qwen3.md)
- [Qwen3-Coder](https://docs.sglang.io/cookbook/autoregressive/Qwen/Qwen3-Coder.md)
- [Qwen3-Coder-Next](https://docs.sglang.io/cookbook/autoregressive/Qwen/Qwen3-Coder-Next.md)
- [Qwen3-Next](https://docs.sglang.io/cookbook/autoregressive/Qwen/Qwen3-Next.md)
- [Qwen3-VL](https://docs.sglang.io/cookbook/autoregressive/Qwen/Qwen3-VL.md)
- [Qwen3.5](https://docs.sglang.io/cookbook/autoregressive/Qwen/Qwen3.5.md)
- [Qwen3.6](https://docs.sglang.io/cookbook/autoregressive/Qwen/Qwen3.6.md)
- [Step-3.7-Flash (new)](https://docs.sglang.io/cookbook/autoregressive/StepFun/Step-3.7-Flash.md)
- [Step3-VL-10B](https://docs.sglang.io/cookbook/autoregressive/StepFun/Step3-VL-10B.md)
- [Step-3.5-Flash](https://docs.sglang.io/cookbook/autoregressive/StepFun/Step3.5.md)
- [Hunyuan 3 Preview](https://docs.sglang.io/cookbook/autoregressive/Tencent/Hunyuan3-Preview.md)
- [Hy3](https://docs.sglang.io/cookbook/autoregressive/Tencent/Hy3.md): Deploy Tencent Hy3 with SGLang — verified launch commands and tuning for the BF16 Mixture-of-Experts model with hybrid thinking, native tool calling, 256K context, and MTP speculative decoding.
- [Inkling](https://docs.sglang.io/cookbook/autoregressive/ThinkingMachines/Inkling.md): Deploy Inkling with SGLang — verified launch commands, tuning, and multimodal / reasoning / tool-calling usage for Thinking Machines' 975B Mixture-of-Experts model with 1M-token context.
- [MiMo-V2-Flash](https://docs.sglang.io/cookbook/autoregressive/Xiaomi/MiMo-V2-Flash.md)
- [MiMo-V2.5](https://docs.sglang.io/cookbook/autoregressive/Xiaomi/MiMo-V2.5.md)
- [Overview](https://docs.sglang.io/cookbook/autoregressive/intro.md): Practical guides for deploying and using large language models and vision language models with SGLang.
- [Autoregressive Model Benchmark Documentation](https://docs.sglang.io/cookbook/base/benchmarks/autoregressive_model_benchmark.md)
- [Diffusion Models Benchmark Documentation](https://docs.sglang.io/cookbook/base/benchmarks/diffusion_model_benchmark.md)
- [Server Arguments](https://docs.sglang.io/cookbook/base/reference/server_arguments.md)
- [Cosmos3](https://docs.sglang.io/cookbook/diffusion/Cosmos/Cosmos3.md)
- [ERNIE-Image](https://docs.sglang.io/cookbook/diffusion/Ernie-Image/Ernie-Image.md)
- [FLUX](https://docs.sglang.io/cookbook/diffusion/FLUX/FLUX.md)
- [Ideogram 4](https://docs.sglang.io/cookbook/diffusion/Ideogram/Ideogram4.md)
- [JoyAI-Echo](https://docs.sglang.io/cookbook/diffusion/JoyEcho/JoyEcho.md): Run JoyAI-Echo multi-shot audio–video generation with SGLang Diffusion.
- [Krea-2](https://docs.sglang.io/cookbook/diffusion/Krea/Krea-2.md)
- [LTX2 & LTX2.3](https://docs.sglang.io/cookbook/diffusion/LTX/LTX2 & LTX2.3.md): Run LTX-2 and LTX-2.3 video generation pipelines with SGLang Diffusion.
- [LingBot World](https://docs.sglang.io/cookbook/diffusion/LingBot-World/LingBot-World.md)
- [LingBot World 2.0](https://docs.sglang.io/cookbook/diffusion/LingBot-World/LingBot-World-2.0.md)
- [LongLive 2.0](https://docs.sglang.io/cookbook/diffusion/LongLive/LongLive-2.0.md): Serve LongLive 2.0 distilled text-to-video and image-to-video models with SGLang-diffusion.
- [MOVA](https://docs.sglang.io/cookbook/diffusion/MOVA/MOVA.md)
- [Qwen-Image](https://docs.sglang.io/cookbook/diffusion/Qwen-Image/Qwen-Image.md)
- [Qwen-Image-Edit-2511](https://docs.sglang.io/cookbook/diffusion/Qwen-Image/Qwen-Image-Edit.md)
- [SANA-WM](https://docs.sglang.io/cookbook/diffusion/SANA-WM/SANA-WM.md)
- [Wan2.1](https://docs.sglang.io/cookbook/diffusion/Wan/Wan2.1.md)
- [Wan2.2](https://docs.sglang.io/cookbook/diffusion/Wan/Wan2.2.md)
- [Z-Image-Turbo](https://docs.sglang.io/cookbook/diffusion/Z-Image/Z-Image-Turbo.md)
- [Overview](https://docs.sglang.io/cookbook/diffusion/intro.md): Practical guides for deploying and using diffusion models with SGLang.
- [SGLang Cookbook](https://docs.sglang.io/cookbook/intro.md)
- [SpecBundle Usage](https://docs.sglang.io/cookbook/specbundle/specbundle_usage.md)
- [Supported Models](https://docs.sglang.io/cookbook/specbundle/supported_models.md)
- [Pi0.5](https://docs.sglang.io/cookbook/vla/OpenPI/Pi0.5.md)
- [Overview](https://docs.sglang.io/cookbook/vla/intro.md): Practical guides for deploying and using Vision-Language-Action policies with SGLang.
- [Adaptive Speculative Decoding](https://docs.sglang.io/docs/advanced_features/adaptive_speculative_decoding.md)
- [Attention Backend](https://docs.sglang.io/docs/advanced_features/attention_backend.md)
- [Breakable CUDA Graph](https://docs.sglang.io/docs/advanced_features/breakable_cuda_graph.md)
- [Checkpoint Engine Integration](https://docs.sglang.io/docs/advanced_features/checkpoint_engine.md)
- [Cuda Graph for Multi-Modal Encoder in SGLang](https://docs.sglang.io/docs/advanced_features/cuda_graph_for_multi_modal_encoder.md)
- [Deterministic Inference](https://docs.sglang.io/docs/advanced_features/deterministic_inference.md)
- [DP, DPA and SGLang DP Router](https://docs.sglang.io/docs/advanced_features/dp_dpa_smg_guide.md)
- [DP for Multi-Modal Encoder in SGLang](https://docs.sglang.io/docs/advanced_features/dp_for_multi_modal_encoder.md)
- [EPD Disaggregation](https://docs.sglang.io/docs/advanced_features/epd_disaggregation.md)
- [Expert Parallelism](https://docs.sglang.io/docs/advanced_features/expert_parallelism.md)
- [Hierarchical KV Caching (HiCache)](https://docs.sglang.io/docs/advanced_features/hicache.md)
- [SGLang HiCache Best Practices](https://docs.sglang.io/docs/advanced_features/hicache_best_practices.md)
- [HiCache System Design and Optimization](https://docs.sglang.io/docs/advanced_features/hicache_design.md)
- [Runtime Attach/Detach HiCache Storage Backend (No Restart)](https://docs.sglang.io/docs/advanced_features/hicache_storage_runtime_attach_detach.md)
- [HiSparse: Hierarchical Sparse Attention](https://docs.sglang.io/docs/advanced_features/hisparse_guide.md)
- [Hyperparameter Tuning](https://docs.sglang.io/docs/advanced_features/hyperparameter_tuning.md)
- [llm-d](https://docs.sglang.io/docs/advanced_features/llm-d.md)
- [LoRA Serving](https://docs.sglang.io/docs/advanced_features/lora.md)
- [Model Loading](https://docs.sglang.io/docs/advanced_features/model_loading.md): Control how SGLang loads model weights: load formats, model loader extra config, multithreaded loading, prefetching, and remote/streaming loaders.
- [Loading Models from Object Storage](https://docs.sglang.io/docs/advanced_features/object_storage.md)
- [Observability](https://docs.sglang.io/docs/advanced_features/observability.md)
- [Advanced Features](https://docs.sglang.io/docs/advanced_features/overview.md): Advanced configuration, optimization, and deployment features for SGLang.
- [PD Disaggregation](https://docs.sglang.io/docs/advanced_features/pd_disaggregation.md)
- [Piecewise CUDA Graph](https://docs.sglang.io/docs/advanced_features/piecewise_cuda_graph.md)
- [Pipeline Parallelism for Long Context](https://docs.sglang.io/docs/advanced_features/pipeline_parallelism.md)
- [Quantization](https://docs.sglang.io/docs/advanced_features/quantization.md)
- [Quantized KV Cache](https://docs.sglang.io/docs/advanced_features/quantized_kv_cache.md)
- [Reasoning Parser](https://docs.sglang.io/docs/advanced_features/separate_reasoning.md)
- [Server Arguments](https://docs.sglang.io/docs/advanced_features/server_arguments.md)
- [SGLang Model Gateway](https://docs.sglang.io/docs/advanced_features/sgl_model_gateway.md)
- [SGLang for RL Systems](https://docs.sglang.io/docs/advanced_features/sglang_for_rl.md)
- [Speculative Decoding](https://docs.sglang.io/docs/advanced_features/speculative_decoding.md)
- [Structured Outputs](https://docs.sglang.io/docs/advanced_features/structured_outputs.md)
- [Structured Outputs For Reasoning Models](https://docs.sglang.io/docs/advanced_features/structured_outputs_for_reasoning_models.md)
- [Tool Parser](https://docs.sglang.io/docs/advanced_features/tool_parser.md)
- [Query VLM with Offline Engine](https://docs.sglang.io/docs/advanced_features/vlm_query.md)
- [Anthropic-Compatible API](https://docs.sglang.io/docs/basic_usage/anthropic_api.md): Use the Anthropic Messages API (/v1/messages) with SGLang, including Claude Code integration and prefix-cache tuning.
- [Amazon SageMaker AI](https://docs.sglang.io/docs/basic_usage/aws_sagemaker.md): Deploy SGLang on Amazon SageMaker AI endpoints using the AWS Deep Learning Container.
- [SGLang Native APIs](https://docs.sglang.io/docs/basic_usage/native_api.md)
- [Offline Engine API](https://docs.sglang.io/docs/basic_usage/offline_engine_api.md)
- [Ollama-Compatible API](https://docs.sglang.io/docs/basic_usage/ollama_api.md)
- [OpenAI-Compatible APIs](https://docs.sglang.io/docs/basic_usage/openai_api.md): Documentation for OpenAI-Compatible APIs
- [OpenAI APIs - Completions](https://docs.sglang.io/docs/basic_usage/openai_api_completions.md)
- [OpenAI APIs - Embedding](https://docs.sglang.io/docs/basic_usage/openai_api_embeddings.md)
- [OpenAI APIs - Vision](https://docs.sglang.io/docs/basic_usage/openai_api_vision.md)
- [Basic Usage](https://docs.sglang.io/docs/basic_usage/overview.md): Core APIs and common usage patterns for SGLang.
- [Sampling Parameters](https://docs.sglang.io/docs/basic_usage/sampling_params.md)
- [Tutorial: Sending a request](https://docs.sglang.io/docs/basic_usage/send_request.md)
- [Bench Serving Guide](https://docs.sglang.io/docs/developer_guide/bench_serving.md)
- [Benchmark and Profiling](https://docs.sglang.io/docs/developer_guide/benchmark_and_profiling.md)
- [Contribution Guide](https://docs.sglang.io/docs/developer_guide/contribution_guide.md)
- [Development Guide Using Docker](https://docs.sglang.io/docs/developer_guide/development_guide_using_docker.md)
- [Development Guide for JIT Kernels](https://docs.sglang.io/docs/developer_guide/development_jit_kernel_guide.md)
- [Evaluating New Models with SGLang](https://docs.sglang.io/docs/developer_guide/evaluating_new_models.md)
- [MSProbe Debugging Guide](https://docs.sglang.io/docs/developer_guide/msprobe_debugging_guide.md)
- [Developer Guide](https://docs.sglang.io/docs/developer_guide/overview.md): Contributing to SGLang — development setup, benchmarking, and evaluation.
- [Quantization Contribution Guide](https://docs.sglang.io/docs/developer_guide/quantization_contribution_guide.md)
- [Installation](https://docs.sglang.io/docs/get-started/install.md): Install SGLang with pip/uv, source, Docker, Kubernetes, and cloud deployment options.
- [Quickstart](https://docs.sglang.io/docs/get-started/quickstart.md): Get up and running with SGLang in minutes: install, launch a server, and send your first request.
- [AMD GPUs](https://docs.sglang.io/docs/hardware-platforms/amd_gpu.md)
- [Apple Silicon with Metal](https://docs.sglang.io/docs/hardware-platforms/apple_metal.md)
- [Contribution Guide](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/ascend_contribution_guide.md)
- [SGLang installation with NPUs support](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/ascend_npu.md): Complete installation guide for SGLang on Ascend NPUs, including component version mapping, environment setup, and launching inference services.
- [Ascend NPU Accuracy Evaluation](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/ascend_npu_accuracy_evaluation.md)
- [Environment Variables](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/ascend_npu_environment_variables.md)
- [Ascend NPU Troubleshooting and FAQ](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/ascend_npu_faq.md)
- [Ascend NPU Operator Development Guide](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/ascend_npu_operator_development.md): How to develop custom operators (Ascend C / Triton) for Ascend NPU and integrate them into the SGLang inference engine.
- [Operator Performance Optimizing Guidance](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/ascend_npu_operator_performance_optimizing.md)
- [Ascend NPU Optimization](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/ascend_npu_optimization.md)
- [Ascend NPU Performance Testing](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/ascend_npu_performance_testing.md)
- [Ascend NPU Performance Profiling Guide](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/ascend_npu_profiling.md)
- [Quantization on Ascend](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/ascend_npu_quantization.md)
- [Ascend NPU Quickstart](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/ascend_npu_quick_start.md)
- [Ascend NPU Reference](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/ascend_npu_reference.md)
- [Ascend NPU Ring-SP Performance (Wan2.1-T2V-1.3B)](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/ascend_npu_ring_sp_performance.md)
- [Support Features on Ascend NPU](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/ascend_npu_support_features.md)
- [Support Models on Ascend NPU](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/ascend_npu_support_models.md)
- [How to Support New Models](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/ascend_npu_support_new_models.md)
- [DeepSeek-R1](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/best_practice/deepseek_r1.md)
- [DeepSeek-V3.2](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/best_practice/deepseek_v3_2.md)
- [GLM-5.1](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/best_practice/glm5_1.md)
- [Kimi-K2.6](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/best_practice/kimi_k2_6.md)
- [MiMo-V2-Flash](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/best_practice/mimo_v2_flash.md)
- [MiniMax-M2.5](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/best_practice/minimax_m2_5.md)
- [Qwen3-8B](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/best_practice/qwen3-8b.md)
- [Qwen3-235B-A22B](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/best_practice/qwen3_235b_a22b.md)
- [Qwen3-30B-A3B](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/best_practice/qwen3_30b_a3b.md)
- [Qwen3-32B](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/best_practice/qwen3_32b.md)
- [Qwen3.5-397B](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/best_practice/qwen3_5_397b.md)
- [Qwen3.6-27B](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/best_practice/qwen3_6_27b.md)
- [Qwen3.6-35B-A3B](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/best_practice/qwen3_6_35b_a3b.md)
- [Qwen3-Next-80B-A3B-Instruct](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/best_practice/qwen3_next_80b_a3b_instruct.md)
- [Mindspore backend](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/mindspore_backend.md)
- [DeepSeek-R1](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/model-tutorials/deepseek_r1.md)
- [DeepSeek-V3.2](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/model-tutorials/deepseek_v3_2.md)
- [GLM-5.1](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/model-tutorials/glm_5_1.md)
- [GLM-5.2](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/model-tutorials/glm_5_2.md)
- [Kimi-K2.6](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/model-tutorials/kimi_k2_6.md)
- [MiMo-V2-Flash](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/model-tutorials/mimo_v2_flash.md)
- [MiniMax-M2.5](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/model-tutorials/minimax_m2_5.md)
- [Qwen3-8B](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/model-tutorials/qwen3-8b.md)
- [Qwen3-235B-A22B](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/model-tutorials/qwen3_235b_a22b.md)
- [Qwen3-30B-A3B](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/model-tutorials/qwen3_30b_a3b.md)
- [Qwen3-32B](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/model-tutorials/qwen3_32b.md)
- [Qwen3.5-397B-A17B](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/model-tutorials/qwen3_5_397b.md)
- [Qwen3.6-27B](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/model-tutorials/qwen3_6_27b.md)
- [Qwen3.6-35B-A3B](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/model-tutorials/qwen3_6_35b_a3b.md)
- [Qwen3-Next-80B-A3B-Instruct](https://docs.sglang.io/docs/hardware-platforms/ascend-npus/model-tutorials/qwen3_next_80b_a3b_instruct.md)
- [CPU Servers](https://docs.sglang.io/docs/hardware-platforms/cpu_server.md)
- [Moore Threads GPUs](https://docs.sglang.io/docs/hardware-platforms/mthreads_gpu.md)
- [NVIDIA GPUs](https://docs.sglang.io/docs/hardware-platforms/nvidia-gpus.md)
- [NVIDIA Jetson Orin](https://docs.sglang.io/docs/hardware-platforms/nvidia_jetson.md): Guide for installing and running SGLang on NVIDIA Jetson Orin devices.
- [Hardware Platforms](https://docs.sglang.io/docs/hardware-platforms/overview.md): Platform-specific guides for running SGLang on GPUs, TPUs, NPUs, CPUs, and more.
- [SGLang Plugin System](https://docs.sglang.io/docs/hardware-platforms/plugin.md)
- [TPU](https://docs.sglang.io/docs/hardware-platforms/tpu.md): SGLang supports high-performance TPU inference through the SGLang-JAX backend, which is specifically optimized for Google Cloud TPUs. The JAX-based implementation delivers exceptional throughput and low latency for Large Language Model (LLM) serving workloads on TPU hardware.
- [XPU](https://docs.sglang.io/docs/hardware-platforms/xpu.md)
- [Custom Chat Template](https://docs.sglang.io/docs/references/custom_chat_template.md)
- [Environment Variables](https://docs.sglang.io/docs/references/environment_variables.md)
- [Troubleshooting and Frequently Asked Questions](https://docs.sglang.io/docs/references/faq.md)
- [Choices Methods in SGLang](https://docs.sglang.io/docs/references/frontend/choices_methods.md)
- [Frontend Language](https://docs.sglang.io/docs/references/frontend/frontend_index.md)
- [SGLang Frontend Language](https://docs.sglang.io/docs/references/frontend/frontend_tutorial.md)
- [Deploy On Kubernetes](https://docs.sglang.io/docs/references/multi_node_deployment/deploy_on_k8s.md)
- [LWS Based PD Deploy](https://docs.sglang.io/docs/references/multi_node_deployment/lws_pd/lws_pd_deploy.md)
- [Multi-Node Deployment](https://docs.sglang.io/docs/references/multi_node_deployment/multi_node.md)
- [Multi-Node Deployment](https://docs.sglang.io/docs/references/multi_node_deployment/multi_node_index.md)
- [DeepSeekV32-Exp RBG Based PD Deploy](https://docs.sglang.io/docs/references/multi_node_deployment/rbg_pd/deepseekv32_pd.md)
- [Nightly precision regression](https://docs.sglang.io/docs/references/nightly_precision_regression.md)
- [References](https://docs.sglang.io/docs/references/overview.md): FAQ, environment variables, production metrics, deployment guides, and more.
- [Post-Training Integration](https://docs.sglang.io/docs/references/post_training_integration.md)
- [Production Metrics](https://docs.sglang.io/docs/references/production_metrics.md)
- [Production Request Tracing](https://docs.sglang.io/docs/references/production_request_trace.md)
- [CLI reference](https://docs.sglang.io/docs/sglang-diffusion/api/cli.md): Run one-off generation tasks and launch the HTTP server from the command line.
- [OpenAI API](https://docs.sglang.io/docs/sglang-diffusion/api/openai_api.md): Image and video generation endpoints with LoRA adapter management.
- [Post-Processing](https://docs.sglang.io/docs/sglang-diffusion/api/post_processing.md)
- [Attention Backends](https://docs.sglang.io/docs/sglang-diffusion/attention_backends.md): Select and configure attention backends for SGLang diffusion pipelines.
- [Cache-DiT Acceleration](https://docs.sglang.io/docs/sglang-diffusion/cache_dit.md): Configure Cache-DiT acceleration for diffusion inference.
- [Caching Acceleration](https://docs.sglang.io/docs/sglang-diffusion/caching-acceleration.md): Compare caching acceleration strategies for diffusion models.
- [CI Performance Baselines](https://docs.sglang.io/docs/sglang-diffusion/ci_perf.md): Generate and update diffusion performance baselines used in CI.
- [Supported Models and Optimization Compatibility](https://docs.sglang.io/docs/sglang-diffusion/compatibility_matrix.md): Check supported SGLang Diffusion models and their optimization compatibility.
- [Contributing to SGLang Diffusion](https://docs.sglang.io/docs/sglang-diffusion/contributing.md)
- [Deployment and Performance Modes](https://docs.sglang.io/docs/sglang-diffusion/deployment_cookbook.md): Choose CPU offload, FSDP, CFG parallelism, SP, TP, and performance-mode presets in SGLang Diffusion.
- [Disaggregated Diffusion Pipeline](https://docs.sglang.io/docs/sglang-diffusion/disaggregation.md)
- [Inference Batching](https://docs.sglang.io/docs/sglang-diffusion/dynamic_batching.md): Batch compatible native SGLang-Diffusion requests during serving.
- [Environment Variables](https://docs.sglang.io/docs/sglang-diffusion/environment_variables.md): Configure SGLang diffusion behavior with environment variables.
- [SGLang Diffusion](https://docs.sglang.io/docs/sglang-diffusion/index.md): Accelerated image and video generation with diffusion models.
- [Install SGLang Diffusion](https://docs.sglang.io/docs/sglang-diffusion/installation.md): Install SGLang Diffusion on NVIDIA, AMD, MUSA, and Ascend platforms.
- [Performance Optimization](https://docs.sglang.io/docs/sglang-diffusion/performance-optimization.md): Choose performance levers for SGLang Diffusion by latency, throughput, memory, and quality tradeoffs.
- [Profiling](https://docs.sglang.io/docs/sglang-diffusion/profiling.md): Profile SGLang diffusion workloads with PyTorch Profiler and Nsight Systems.
- [Progressive Resolution Generation](https://docs.sglang.io/docs/sglang-diffusion/progressive_resolution.md): Experimental spectral progressive resolution growing for selected SGLang Diffusion pipelines.
- [Quantization](https://docs.sglang.io/docs/sglang-diffusion/quantization.md)
- [Sequence Parallelism](https://docs.sglang.io/docs/sglang-diffusion/ring_sp_performance.md)
- [Support New Diffusion Models](https://docs.sglang.io/docs/sglang-diffusion/support_new_models.md): A concise implementation guide for adding diffusion model families to SGLang-Diffusion.
- [TeaCache Acceleration](https://docs.sglang.io/docs/sglang-diffusion/teacache.md): Configure TeaCache for temporal similarity-based diffusion acceleration.
- [Supported models](https://docs.sglang.io/docs/supported-models.md): See which families of SGLang-compatible models are actively maintained.
- [Classification Models](https://docs.sglang.io/docs/supported-models/classify_models.md)
- [Diffusion language models](https://docs.sglang.io/docs/supported-models/diffusion_language_models.md)
- [Embedding models](https://docs.sglang.io/docs/supported-models/embedding_models.md): Dense and sparse embedding models with FlashInfer acceleration and SGLang's batching infrastructure.
- [Large Language Models](https://docs.sglang.io/docs/supported-models/generative_models.md)
- [MindSpore Models](https://docs.sglang.io/docs/supported-models/mindspore_models.md)
- [Use Models From ModelScope](https://docs.sglang.io/docs/supported-models/modelscope.md)
- [Multimodal Language Models](https://docs.sglang.io/docs/supported-models/multimodal_language_models.md)
- [Rerank models](https://docs.sglang.io/docs/supported-models/rerank_models.md)
- [Reward models](https://docs.sglang.io/docs/supported-models/reward_models.md)
- [How to Support New Models](https://docs.sglang.io/docs/supported-models/support_new_models.md): This document explains how to add support for new language models and multimodal large language models (MLLMs) in SGLang. It also covers how to test new models and register external implementations.
- [Transformers Fallback in SGLang](https://docs.sglang.io/docs/supported-models/transformers_fallback.md)
- [Welcome to SGLang](https://docs.sglang.io/index.md): High-performance serving framework for large language and multimodal models.