洞察 AI 业界
最新进展

智能指数 29.5 · 输出吞吐 175.63 tok/s

智能指数 37.8 · 输出吞吐 178.92 tok/s

智能指数 43.4 · 输出吞吐 235 tok/s

智能指数 41.2 · 输出吞吐 192.03 tok/s

智能指数 34.5 · 输出吞吐 175.46 tok/s

智能指数 38.4 · 输出吞吐 - tok/s

智能指数 41.1 · 输出吞吐 213.13 tok/s

智能指数 52.6 · 输出吞吐 - tok/s

智能指数 42.1 · 输出吞吐 54.14 tok/s

智能指数 47.8 · 输出吞吐 55.37 tok/s

智能指数 51 · 输出吞吐 60.37 tok/s

智能指数 50.2 · 输出吞吐 58.93 tok/s

# Release v5.19.0 ## New Model additions ### EmbeddingGemma2 <img width="2716" height="2308" alt="image" src="https://github.com/user-attachments/assets/84734e74-163d-4d12-b166-ffcf6749d563" /> EmbeddingGemma 2 is a multimodal embedding model from Google built on the Gemma 4 architecture. It encodes text, images, audio, and video, individually or combined in one input, into a shared 768-dimensional vector space for cross-modal retrieval, semantic similarity, clustering, and classif

# v0.31.0 ## Highlights This release features 717 commits from 307 contributors (96 new)! * **DeepSeek-V4.1-Flash performance**: FlashMLA mega attention with the V4.1 NVFP4 compressed KV cache is now the SM100 default (#56935); DeepGEMM sparse MQA logits for the indexer (#56254) and Mega-Gate fusing the gate GEMM with expert selection (#56266); decoder boundaries fuse the TP all-reduce, mHC input preparation (#57643) and the MoE finalize (#58586); a fused small-batch WO-A with inverse R

# Highlights *779 PRs from 227 contributors.* ## New models | Model | Type | Cookbook | |---|---|---| | DeepSeek-V4.1 Flash | LLM / VLM | [link](https://docs.sglang.io/cookbook/autoregressive/DeepSeek/DeepSeek-V4_1) | | GigaChat 3.5 | LLM / VLM | [link](https://docs.sglang.io/cookbook/autoregressive/GigaChat/GigaChat3.5) | | IQuest-Q1 | LLM / VLM | [link](https://docs.sglang.io/cookbook/autoregressive/IQuestLab/IQuest-Q1) | | MiMo-V2.6 / MiMo-V2.6-Pro | LLM / VLM | [link](https://d

## New Model additions ### Nemotron 3 Diarization <img width="1680" height="900" alt="image" src="https://github.com/user-attachments/assets/fe735cb3-9e5b-43ad-8f60-9dec8425aec7" /> Nemotron 3 Diarization is an open-weight streaming speaker diarization model designed to determine "who spoke when" in real-world audio. It supports both streaming and offline inference, handles up to eight speakers, and orders speaker outputs by each speaker's first arrival in the input audio. The mode

# v0.30.0 ## Highlights This release features 762 commits from 315 contributors (104 new)! * **New models**: DeepSeek-V4.1-Flash (#56214, #56228, #56208) with the whole KV stored in MXFP8 through the FlashMLA V4.1 record on SM100 (#56893), DeepGEMM Mega-mHC (#56962), and async Engram prefetch with Engram DP sharding (#56512); DeepSeek-V4-Flash-Vision-Exp (#54566), also on ROCm (#55107) and with LoRA (#55897); GLM-5.3-Flash (#53906) with EPLB (#55119); K2-Horizon (#55063); Cohere Compass

# Highlights *713 PRs from 237 contributors.* **New models in this release** (see the [cookbook](https://docs.sglang.io/cookbook) for all supported models): | Model | Type | PRs | Cookbook | |---|---|---|---| | GLM-5.3-Flash | Autoregressive | [#36507](https://github.com/sgl-project/sglang/pull/36507), [#38621](https://github.com/sgl-project/sglang/pull/38621) | [link](https://docs.sglang.io/cookbook/autoregressive/GLM/GLM-5.3-Flash) | | Hy4-Preview | Autoregressive | [#36805](https:

# Release v5.17.0 ## New Model additions ### HYV4 <img width="1503" height="827" alt="image" src="https://github.com/user-attachments/assets/e6ed85ee-eb1d-40eb-a0d4-c649f6337ca9" /> Hy4-Preview is a 780B-parameter mixture-of-experts language model that activates 49B parameters per token. Each MoE layer holds 256 routed experts plus one always-active shared expert and routes every token to 8 of them. The context window is 1M tokens. The architecture combines four features:

# v0.29.0 ## Highlights This release features 594 commits from 277 contributors (91 new)! * **Model Runner V2 is now the default for all models** (#53183), completing the rollout that began with pooling models (#48290). MRV2 also gained CUDA graph memory profiling for KV cache auto-sizing (#53306), batch-sharded sampling that cuts per-step logits memory by 1/TP (#50465), prompt embeds (#42963), `extract_hidden_states` speculation (#49811), padded FULL cudagraph dispatch for uniform deco

# Highlights *786 PRs from 214 contributors.* **New models in this release** (see the [cookbook](https://docs.sglang.io/cookbook) for all supported models): | Model | Type | PRs | Cookbook | |---|---|---|---| | Qwen3.8 (2.4T-A95B) | Autoregressive | [#35758](https://github.com/sgl-project/sglang/pull/35758) ⭐ | [link](https://docs.sglang.io/cookbook/autoregressive/Qwen/Qwen3.8) | | Qwen3.8-27B | Autoregressive | [#34859](https://github.com/sgl-project/sglang/pull/34859) | [link](http

# Release v5.16.1 This is a special release as we include GLM! (and a few small fixes) # GLM-5.3-Flash <img width="4239" height="2643" alt="image" src="https://github.com/user-attachments/assets/17bc9c29-758b-44c8-8230-42f945ded209" /> GLM-5.3-Flash, the first **natively multimodal model** in the GLM-5 series. With 320B total parameters and just 18B active parameters, it outperforms GLM-5.2 across benchmarks and real-world workloads at one-tenth the price, while approaching Claude Op

# Release v5.16.0 ## New Model additions ### Qwen4-Exp <img width="2241" height="693" alt="image" src="https://github.com/user-attachments/assets/c838b5ba-ffea-42da-baa9-3f66178e3671" /> Qwen4-Exp builds on Qwen3.5's hybrid text and multimodal architecture with three key components: GatedResidual (GR), Qwen Sparse Attention (QSA), and Per-Layer Embedding (PLE). GR is a Qwen-developed residual architecture that combines Hyper-Connection with GatedNorm. It mixes multiple residual

# v0.28.0 ## Highlights This release features 584 commits from 270 contributors (76 new)! * **Kimi-K3 performance push**: a major optimization effort for Kimi-K3 across the stack — Decode Context Parallel (DCP) support (#50484), fused FlashKDA decode and prefill kernels (#50654, #51311, #52458), SiTU activation support for MegaMoE (#50510), GEMM-RS for sequence parallelism (#52079), combined all-gathers with 1.5~3x kernel-level speedup (#51070), an adaptive speculative token budget deli

Recursive self-improvement (RSI) aims to transform interaction experience into persistent improvements, yet external feedback is costly and world-model predictions may become unreliable as policies evolve. We introduce RSI-Anything, a domain-general framework built on JEPA-Anything that uses predictive world models to guide action and learning-data selection. Observed feedback drives policy or workflow improvement and conditional world-model adaptation, while component and compatibility checks g

Vision-Language Models (VLMs) increasingly analyze and reason about images, yet it remains unclear whether their internal representations faithfully encode the structured relations required for such reasoning. We introduce EICS (Evaluation-Invariant Concept Stability), a metric for testing whether representations consistently encode concepts under controlled one-concept interventions, and CASP-VL (Causally controlled Spatial Perturbations for Vision-Language models), a benchmark of causally cont

Neural semantic parsers generate meaning representations token by token, yet well-formedness depends on references, scope, and graph relations that unfold during generation. Constrained decoding prevents locally invalid tokens but does not guarantee that a valid prefix can be completed within a fixed output-length limit. We investigate structural automata at three stages of DRS parsing (decoding, post hoc processing, and training) and additionally evaluate decoding on AMR. For decoding, we intro

Large language models reuse the same attention heads across tasks, but identifying these shared heads does not explain what they do differently in each task. A natural source of inspiration is task-dependent interactions among shared brain regions, which motivate a network view of language-model computation. Here, we compare fixed model components across tasks using identical graph facts, statements specifying which directed connections are present. We measure how interventions affect attention

Comics are one of the great visual records of the twentieth century: hundreds of thousands of pages in which a mass readership saw its leaders, its wars and its heroes. Most of this record has never been studied as a whole; comics scholarship proceeds one book at a time. We argue that artificial intelligence, including generative models, can change this to the benefit of the medium itself. We propose PAGE, a framework that makes artificial intelligence an instrument for comics research. It parse

Recent advances in automatic pronunciation assessment (APA) have driven a shift from scripted to free-form assessment. However, existing methods often rely on multi-stage pipelines with forced alignment and separate feature extractors, limiting their practicality across response settings. In light of this, we propose OpenPA, a novel APA framework that leverages Whisper's cross-attention mechanism to jointly predict word- and utterance-level pronunciation scores without forced alignment. Through

A useful language model should know when to answer and when to say "I don't know." Can we improve that decision by training the model itself or by training a separate component to check its answers? We study both approaches on Qwen3-14B for factual questions, using a score that rewards correct answers, penalizes mistakes, and allows abstention. Across three training seeds and two held-out confidence thresholds, a learned answer selector improves threshold-averaged utility over the original model

Large language models (LLMs) have shown strong results in classical code generation and are increasingly used to write the unit tests that safeguard that code. Whether they can do the same for quantum software is less clear, because quantum programs are probabilistic, depend on numerical tolerances, and are judged by domain-specific checkers, so a well-formed test can still disagree with the benchmark's verdict. Existing quantum benchmarks measure whether a model produces a correct solution, but

Large language models (LLMs) are now a standard source of query expansion terms but they hallucinate. They inject non existent entities and off-topic terms that add ranking noise and can hurt strong retrievers. Existing remedies such as corpus grounding, mutual verification and preference alignment treat hallucinated terms as an undifferentiated nuisance to be filtered by agreement heuristics, without a mechanistic account of why a given term appears or a principled criterion for how many terms

Data augmentation is standard practice for low-resource automatic speech recognition (ASR), but its benefits are almost always reported as aggregate error rates, leaving open why and when a method helps. On four low-resource languages and English at five data sizes, we cross two augmentations that modify existing audio (speed perturbation and SpecAugment) with two that modify text transcriptions and synthesize audio from it. We analyze errors at the phoneme level and find that they help for comp

Detecting communicative meaning in dialogue requires modeling not only lexical content but also the participant-role structures through which sentiment, emotion, politeness, and formality (SEPF) are expressed. Existing approaches rely on contextual language representations while leaving semantic roles implicit. We introduce a role-aware transformer architecture that maps the four SEPF dimensions to theory-grounded Semantic Role Labeling (SRL) configurations, incorporating these configurations di

Discourse particles subtly shape the flow of discourse by reflecting conversational context and interlocutor attitudes, often leaving the truth-conditional meaning untouched. They are also often polyfunctional, with distinct senses which partially overlap with other particles; for example, the exclusive use of English "just" and "only" ("John only/just drinks water"). Despite the frequency of these words in natural language, they have gone largely unstudied due to the subtle nature of their effe