The recent SGLang × MUSA meetup revealed that MUSA’s GPU backend has been merged into SGLang’s official codebase, delivering zero‑learning‑cost integration, performance gains of up to 66 % on DeepSeek‑V4, and a growing ecosystem of adapters, high‑performance kernels, and distributed inference support.
Ant Group's Ling‑2.6‑1T, a 1‑trillion‑parameter LLM built for token efficiency and fast‑thinking, outperforms on elite reasoning and agentic benchmarks, offers easy local deployment via vLLM or SGLang, provides a quantized 3.6‑bit version, and includes practical usage tips for developers and knowledge workers.
RadixArk raised a $100 million seed round backed by top hardware and AI investors to turn the open‑source SGLang inference engine and the Miles RL framework into day‑0 standards, aiming to democratize AI infrastructure and eliminate bottlenecks from training to inference.
The article analyzes DeepSeek‑V4's architectural innovations—including mixed sparse attention, mHC, and native FP4 weights—explains SGLang's ShadowRadix, HiSparse, and in‑graph speculative decoding solutions, presents benchmark gains, provides Docker deployment steps, and warns of key pitfalls for long‑context inference.
Kimi K2.6, an open-source 1-trillion-parameter MoE model, expands Agent capabilities with 256K context, multimodal inputs, and the ability to coordinate 300 sub-Agents over 4,000 steps, achieving top scores on benchmarks like Terminal-Bench 2.0, SWE-Bench Pro, and BrowseComp, while offering flexible deployment via vLLM, SGLang, and KTransformers.
The DFlash approach replaces speculative decoding’s autoregressive drafter with a block diffusion model and injects target‑model hidden features into every KV‑cache layer, achieving up to 5× speed‑up for Qwen3.5‑27B on single‑GPU and 1.5–1.9× on high‑concurrency workloads while preserving output quality.
MiniMax‑M2.7, the newly open‑sourced 230‑billion‑parameter MoE model, offers self‑evolution, professional software engineering and agent capabilities, and can be deployed locally using Ollama, vLLM, SGLang or Docker with 4‑8 H200 GPUs, while the article details hardware needs, performance gains and tool‑calling/Thinking features.
This article reviews the DeepLearning.ai short course on SGLang, explains why large‑language‑model inference is slow, details how KV Cache reduces the computation from O(n²) to O(n), introduces RadixAttention for cross‑request caching, and presents code examples and benchmark results showing up to 10× speedup in real‑world RAG scenarios.
The article examines the hype around “harness engineering” in LLM workflows, showing through SGLang’s multi‑agent experience that the approach merely repackages established software‑engineering principles such as separation of concerns, docs‑as‑code, and structured routing, and discusses its limits and future implications.
The article traces five eras of KV cache management for LLM inference—from its absence before Transformers to the emerging unified hybrid memory architecture—comparing vLLM, SGLang, and TensorRT‑LLM and offering a decision framework for selecting the right solution in various deployment scenarios.
This guide walks through preparing the Kimi 2.5 model, uploading it to OSS, configuring persistent storage, and using SGLang, RoleBasedGroup, and Openclaw to deploy a production‑grade inference service on Alibaba Cloud Kubernetes with step‑by‑step commands and YAML examples.
GLM‑ASR‑Nano‑2512, a 1.5 B‑parameter open‑source speech‑recognition model released in December 2025, delivers state‑of‑the‑art accuracy on Chinese dialects and low‑volume audio, outperforms Whisper V3 on benchmark tests, runs on consumer GPUs, and provides detailed installation and deployment guides for transformers, vLLM and SGLang.
SGLang, the open‑source inference accelerator, has been spun out into RadixArk—a $400 million‑valued startup aiming to democratize AI infrastructure, while the broader market sees a surge of funding for inference‑focused companies.
This guide details how switching from vLLM to SGLang on eight NVIDIA H800 GPUs increased Llama‑3‑70B‑Instruct throughput from 180 to 420 tokens per second, covering SGLang’s core innovations, environment setup, configuration tweaks, performance benchmarks, troubleshooting tips, and production‑grade deployment scripts.
This article explains how SGLang’s hybrid model design combines Transformer attention with Mamba state‑space layers, introduces a dual‑pool memory architecture and elastic allocation, and presents specialized prefix‑cache and speculative‑decoding techniques that together enable efficient, scalable inference for long‑context large language models.
The article details the Attention‑FFN Disaggregation (AFD) technique used by Baidu Baige to separate self‑attention and feed‑forward network stages in DeepSeek‑V3 models, describing multi‑stage scheduling, three‑batch overlap, communication optimizations, and performance results that achieve up to 19% throughput improvement under a 100 ms SLO.
Token‑level Two‑Chunk Overlap replaces traditional batch‑level Two‑Batch Overlap, dynamically splitting sequences into balanced token chunks, enabling near‑equal compute and communication times, improving GPU utilization and achieving up to 30% throughput gains in heterogeneous request workloads, with zero accuracy loss.
LongCat-Flash, an open‑source Mixture‑of‑Experts model released by Meituan, leverages model‑system co‑design, PD‑disaggregation, SBO scheduling and large‑scale expert parallelism within the SGLang framework to deliver dramatically lower latency, higher throughput and cost‑effective inference for AI agents, with detailed deployment instructions provided.
This article examines how Volcano Engine's Elastic Instant Cache (EIC) tackles the memory bottleneck, high‑concurrency latency, and cross‑node coordination challenges of large language model inference by decoupling storage and computation, pooling resources, and applying layered optimizations, ultimately boosting AI inference efficiency, scalability, and cost‑effectiveness across various deployment scenarios.
This article shares a detailed, first‑hand analysis of the technical challenges, framework choices, memory management, weight conversion, precision alignment, and efficiency optimizations encountered while building reinforcement‑learning pipelines for a 671‑billion‑parameter model using the verl ecosystem.
This article provides a comprehensive comparison of seven popular large‑language‑model inference engines—Transformers, vLLM, Llama.cpp, SGLang, MLX, Ollama and others—detailing their core features, performance characteristics, hardware compatibility, concurrency support, and ideal use‑cases, plus practical installation guidance for Xinference.
This guide shows how to use the lightweight OpenStation platform to install, configure, and launch the DeepSeek‑R1 large‑model on a personal server in under 15 minutes, covering zero‑code deployment, resource management, inference engine selection, and integration with CherryStudio.
The article summarizes Zhihu’s technical talk on the ZhiLight large‑model inference framework, detailing model execution mechanisms, GPU load analysis, multi‑GPU parallel strategies, open‑source engine comparisons, compute‑communication overlap, quantization techniques, benchmark results, and future directions for scalable LLM deployment.
Meituan’s search and recommendation team converted the FP8‑only DeepSeek‑R1 model to INT8 by first casting weights to BF16 and then applying block‑wise or channel‑wise quantization, which preserves GSM8K and MMLU accuracy while delivering 33% to 50% higher throughput on A100‑80G GPUs, and they released the SGLang‑based inference scripts and quantized weights publicly, enabling deployment on older NVIDIA hardware without accuracy loss.
The Inspur Metabrain R1 inference server, equipped with FP8 acceleration and a 1128 GB HBM3e memory pool, has been tightly integrated with SGLang 0.4.3 to run the 671‑billion‑parameter DeepSeek R1 model, delivering over 1,000 concurrent user sessions and up to 3,976 tokens/s throughput.
