GPTProto
⌘K

Qwen / Alibaba News

Latest reporting, research and product updates filed under Qwen / Alibaba.

76 picksNewest firstLatest article Sep 23, 2026, 5:11 PM

Latest stories

2 stories
MarkTechPost（RSS）AI score 61/100

Cloudflare Releases Clef and Clef-flash: Open-Weight Decision Models That Return Typed Probabilities Instead of Text

Cloudflare has released Clef and Clef-flash, the first models trained by its Workers AI team. They are decision models, not chatbots. Each reads an input state and a schema of typed questions. It returns a probability for every allowed answer, with no free-form text. Both are open-weight under Apache 2.0 and compatible with TypeSafe AI’s Jev API. Is it deployable? Yes, Both models run today on Workers AI, and the weights are on Hugging Face for self-hosting. What a Decision Model Does An LLM gen
2 stories
HuggingFace Daily Papers（社区热门论文）AI score 31/100

Decision-Oriented Recommendation Reranking: An Empirical Study of Jev

View PDF HTML (experimental) Abstract:Large language models (LLMs) have shown promise for recommendation reranking, but their use introduces an important tradeoff between recommendation quality and serving efficiency. We investigate whether a decision-oriented model provides a useful alternative when the reranking task is fundamentally a structured choice among predefined candidate items. Specifically, we conduct a controlled empirical study of Jev, described by TypeSafe AI as a ``System One Mod
HuggingFace Daily Papers（社区热门论文）AI score 45/100

PivotOPD: Learning to Recover from Pivotal Mistakes in Multi-Turn Agents

View PDF HTML (experimental) Abstract:On-policy distillation (OPD) is a promising approach for training language agents, providing dense teacher supervision on student-generated trajectories. However, in multi-turn interaction, an incorrect action changes the states the student encounters later, so errors compound across turns. In preliminary experiments across three Qwen3 models (8B to 235B), we find that more than half of the failed rollouts contain a pivotal mistake, an action that moves the

2 stories
MarkTechPost（RSS）AI score 72/100

Alibaba Qwen Releases Qwen-Audio-3.1-Realtime: A Full-Duplex Voice Model Trained to Think, Act, and Decide When to Speak

Alibaba’s Qwen team has released Qwen-Audio-3.1, a 5-model audio stack spanning ASR, TTS and realtime interaction. The main model is Qwen-Audio-3.1-Realtime, a full-duplex speech model built for voice agents that call tools. Qwen also cut prices: about 85% on Realtime, about 70% on TTS and up to 95% on ASR. Is it deployable? Yes, as a managed API. qwen-audio-3.1-realtime-plus is live on QwenCloud over WebSocket. No open weights were announced. What Ships on QwenCloud The model page lists text an
HuggingFace Daily Papers（社区热门论文）AI score 40/100

AREX-2: Advancing Self-Improving Agents through Long-Horizon Reflective Tasks

Authors:Hongjin Qian, Chaofan Li, Kun Luo, Wenqing Wei, Jianlyu Chen, Shuqi Lu, Yuyang Hu, Hongwang Xiao, Hui Wang, Chaozhuo Li, Qiwei Ye, Zhicheng Dou, Defu Lian, Zheng Liu View PDF HTML (experimental) Abstract:We present AREX-2, an effort to advance the self-improving capability of LLM agents, which we define as the ability to iteratively refine a solution at test time. This ability rests on two complementary capabilities: reflection, which produces a solution better than the current one, and

9 stories
HuggingFace Daily Papers（社区热门论文）AI score 39/100

Rubric Rewards from Item Response Theory

View PDF HTML (experimental) Abstract:Many language tasks have no single answer that can be checked automatically. Rubrics provide criteria for judging responses to these tasks. For reinforcement learning, the resulting verdicts must be combined into a scalar reward. A common approach sums the points assigned to satisfied criteria. Distinct verdict patterns can thus receive the same reward, and the fixed points encode how much each criterion should count, not how strongly its verdict distinguish
HuggingFace Daily Papers（社区热门论文）AI score 38/100

Rethinking Latent Visual Reasoning: Grounding Latent Reasoning in Visual Evidence

Authors:Xi Xiao, Tianchen Zhao, Youngeun Kim, Zhuowei Li, Linghan Xu, Jiaye Wu, Zheng Zhang, Xiang Xu, Xuanbai Chen, Farhan Tejani, Jakub Zablocki, Julia Xu, Yifan Xing View PDF HTML (experimental) Abstract:Latent visual reasoning (LVR) enables multimodal large language models (MLLMs) to perform intermediate computation in continuous latent tokens rather than expressing every reasoning step in words. However, unlike textual CoT, latent reasoning is not directly observable, making it difficult to
HuggingFace Daily Papers（社区热门论文）AI score 39/100

Label-free steering: Compressing test-time reinforcement learning into bias-only subspaces

View PDF HTML (experimental) Abstract:Test-time reinforcement learning (TTRL) enables models to improve their reasoning without relying on labeled training data, but existing approaches typically optimize a large fraction of the model parameters. This raises a natural question: can effective test-time adaptation emerge when both the reward signal and the optimization space are severely restricted? We answer this question with label-free bias-only TTRL, which uses majority-vote pseudolabels as re
HuggingFace Daily Papers（社区热门论文）AI score 56/100

Same Bytes, Different Authority: Reserved-Token Representations in Chat-Template Prompt Injection

View PDF HTML (experimental) Abstract:Prompt injection against LLM agents becomes much stronger when the injected instruction is wrapped in the model's own chat template. A forged template marker such as <|im_start|> can reach the model either as a single reserved control token or as a sequence of ordinary subword tokens. The two decode to exactly the same text, and because tokenization runs on the server, the defender rather than the attacker decides which one the model receives. We use this to
HuggingFace Daily Papers（社区热门论文）AI score 62/100

Language Models Are "Insecure" Reporters

View PDF HTML (experimental) Abstract:As large language models are deployed in increasingly autonomous long-horizon tasks, manually auditing and verifying the actions, artifacts, and outputs of models becomes more difficult. Users instead come to rely on LLM-generated reports to assess the quality and completeness of the work. We introduce a suite of eight adversarial reporting scenarios to systematically study whether LLMs conceal narrative-changing flaws: errors or limitations that undermine a
HuggingFace Daily Papers（社区热门论文）AI score 32/100

PMOPD: Task Ordering, Cycling, and Parameter-Update Subspace Protection in Multi-Teacher On-Policy Distillation

View PDF HTML (experimental) Abstract:Multi-teacher on-policy distillation (MOPD) has emerged as a popular post-training paradigm for integrating specialized capabilities in frontier language models. Existing OPD research has primarily focused on optimizing single-task distillation through objective design, distillation scope, and teacher signal construction, whereas MOPD must aggregate multiple capabilities in shared parameters and address the resulting capability seesaw, in which improving one
HuggingFace Daily Papers（社区热门论文）AI score 40/100

See it, Say it, Sorted: Mechanistic Diagnosis and Parameter-Space Mitigation of Emergent Misalignment in LLMs

View PDF HTML (experimental) Abstract:Safety-aligned LLMs can exhibit emergent misalignment (EM): narrow domain adaptation unexpectedly triggers catastrophic safety failures across unrelated domains. Prior static analyses leave training dynamics unmapped, while existing defenses rely on heuristics that degrade utility. We present a dynamic, second-order geometric study of EM. Tracking training trajectories reveals that directional Hessian curvature concentrates sharply on semantic pivot tokens.

HuggingFace Daily Papers（社区热门论文）AI score 42/100

LLMs are General Asynchronous Agents

View PDF HTML (experimental) Abstract:Modern LLMs are increasingly capable as autonomous agents, but they follow sequential interaction cycles: read, think, reply or call tools, repeat. Many real-world use cases are not sequential: voice assistants, embodied agents, and monitoring systems receive new inputs while they think or perform another task. Modern LLMs address this with specialized architectures for voice interaction and video streams, VLAs for robot control, asynchronous tool calling fo
HuggingFace Daily Papers（社区热门论文）AI score 44/100

ROSS: Relearning from Self-Generated Rollouts through Selective Supervision

View PDF HTML (experimental) Abstract:Large language model post-training generates self-generated rollouts through reinforcement learning and on-policy distillation, yet this experience is often treated as stale once the policy advances. Historical rollouts can remain compatible with a later policy while preserving behaviors that the policy no longer expresses reliably. However, they may also contain mistakes, abandoned attempts, and redundant actions that should not be imitated, motivating fine
1 story
HuggingFace Daily Papers（社区热门论文）AI score 40/100

QwenGyre: An Elastic Reinforcement Learning Framework for Training xLong-Horizon Agents

Authors:Weiqi Wang, Yuxin Zhou, Mouxiang Chen, Siyuan Zhang, Yi Zhang, Yuyan Luo, Zhiyu Yin, Chencan Wu, Jiemin Jiang, Wentao Yao, Chujie Zheng, JianWei Zhang View PDF HTML (experimental) Abstract:Large language model (LLM) agents increasingly undertake extreme-long (xlong) horizon tasks, where a single execution can span hours, hundreds of model--environment interactions, and nearly 1M tokens per rollout. Applying online reinforcement learning (RL) to such executions poses two fundamental chall
1 story
HuggingFace Daily Papers（社区热门论文）AI score 39/100

RoboFoundry: System-as-Policy Evolution for Self-Learning Embodied Agents

Authors:Jingsong Liang, Shuhao Liao, Shizhe Zhang, Diyuan Hou, Yuxin Cai, Xinjian Deng, Chengyang He, Wenhui Huang, Runjia Tan, Zhidong Wang, Lan Yu, Xuesong Tian, Guillaume Sartoretti, Jie Luo, Yao Mu, Wenjun Wu, Wanhua Li, Chen Lv View PDF HTML (experimental) Abstract:A foundation model should not act in isolation as an embodied agent. Yet, existing methods often optimize individual components of the agent stack, such as memory, context, skills, or action interfaces, rather than treating the s
1 story
1 story
1 story
1 story
1 story
1 story
2 stories
1 story
1 story
1 story
1 story
Hugging Face：Blog（RSS）AI score 66/100

Accelerating Transformers Fine-Tuning with NVIDIA NeMo AutoModel

NVIDIA NeMo AutoModel 是基于 Transformers v5 的开源库，添加 Expert Parallelism、DeepEP 融合 all-to-all 调度和 TransformerEngine 内核。在 MoE 模型微调中，相比原生 v5，训练吞吐量提升 3.4–3.7 倍，GPU 内存减少 29–32%，仅需改动一行 import。在 16 节点 128 张 H100 上全微调 Nemotron 3 Ultra 550B A55B 时，v5 因内存不足无法运行，而 AutoModel 凭借 EP=64 专家并行使训练可行。单节点 30B MoE 模型（如 Qwen3-30B-A3B）同样获得可量化的性能优势。
1 story
LMSYS：Blog（Chatbot Arena 团队）AI score 67/100

Blog MOSS-TTS Local Transformer v1.5 on SGLang-Omni: Serving Native-Streaming 48 kHz Speech Today we are announcing end-to-end serving for MOSS-TTS-Local-Transformer-v1.5 on SGLang-Omni, together with MOSI and the OpenMOSS Team. MOSS-TTS-Local-Transformer-v1.5 is an open TTS model for 48 kH... MOSI, OpenMOSS Team & SGLang-Omni Team

MOSS-TTS-Local-Transformer-v1.5 是一款开源 TTS 模型，支持 48 kHz 立体声、零样本声音克隆、最长 10 分钟长文本合成、时长控制及 31 种语言。其核心采用 Qwen3-4B 骨干与约 2B 参数的 MOSS-Audio-Tokenizer-v2 音频编解码器，通过 12 个 RVQ 码本运行。SGLang-Omni 以三阶段流水线部署该模型。在 Seed-TTS-Eval 上词错误率 5.10%、语音相似度 69.23%，CV3-Eval 上 WER 7.48%、SIM 61.59%，MiniMax Multilingual 上 WER 6.37%、SIM 75.31%，X Voice 上 WER 20.48%、SIM 63.00%。
1 story
LMSYS：Blog（Chatbot Arena 团队）AI score 67/100

Blog The next generation of speculative decoding: DFlash and Spec V2 Using Modal and Z Lab's DFlash speculative decoding models with SGLang’s newly default Spec V2 engine, you can achieve state-of-the-art latencies for LLM inference serving. Our new, jointly-released D... Z Lab, Modal, and SGLang Teams

Z Lab、Modal 与 SGLang 团队联合发布 DFlash 投机解码模型和 SGLang 的默认 Spec V2 引擎。DFlash 采用块扩散+KV 注入并行生成整块 draft token，在 Qwen 3.5 397B-A17B（BF16）的 HumanEval 数据集上、并发 1 时吞吐量达到基线的 4.3
1 story
蚂蚁 inclusionAI：HuggingFace 新模型AI score 71/100

inclusionAI/VISTA-9B

VISTA-9B是基于Qwen3.5 9B骨干训练的GUI定位模型，输入截图与自然语言指令，输出0-1000归一化坐标。采用VISTA（视图一致自验证）方法，含view-consistent GRPO与self-verified cross-view anchoring。在SSPro、SSV2、OSWorld-G、OSWorld-G-R上分别取得69.2、95.8、68.1、75.5分，超越Qwen3.5-9B与GRPO-9B基线。模型已开源，可通过HuggingFace加载使用。
1 story
1 story
X：通义千问 / Qwen (@Alibaba_Qwen)AI score 69/100

Fast, faster, Qwen. 🚀 Thrilled to see Qwen3.5 reaching a record-breaking 580 tps for agentic workl...

Qwen3.5在TokenSpeed推理引擎上，针对智能体工作负载达到了创纪录的580 tokens per second (tps)速度。这一成果由通义千问推理团队、lightseekorg Foundation TokenSpeed团队、NVIDIA及Mooncake团队共同实现，并采用了tri_dao的FlashAttention-4 (FA4) 优化。此里程碑标志着开源大语言模型推理性能的边界得到了推动，相关详情可查阅PyTorch社区博客。
1 story
Hugging Face：Blog（RSS）AI score 63/100

Towards Speed-of-Light Text Generation with Nemotron-Labs Diffusion Language Models

NVIDIA 发布 Nemotron-Labs Diffusion 系列，含 3B、8B、14B 文本模型和 8B 视觉-语言模型（VLM），均采用商用友好的 NVIDIA Nemotron Open Model License 或 NVIDIA Source Code License。模型支持自回归、扩散（逐块并行生成后逐步精炼）和自推测（扩散草拟候选 token 再自回归验证）三种模式。8B 模型平均准确率比 Qwen3 8B 提升 1.2%，扩散模式每次前向传递的 token 数（TPF）达自回归的 2.6 倍，自推测达 6–6.4 倍。模型在 1.3T tokens 上预训练、45B tokens 上微调，代码与模型已发布于 HuggingFace 和 GitHub，推理将获 SGLang 支持。
1 story
Hugging Face：Blog（RSS）AI score 58/100

MedQA: Fine-Tuning a Clinical AI on AMD ROCm — No CUDA Required

该项目使用AMD Instinct MI300X（192 GB HBM3显存）和ROCm，通过LoRA微调Qwen3-1.7B模型实现医学问答。训练仅用2000条MedMCQA样本，约5分钟完成，仅更新约220万参数（占模型总参数的0.1443%），全程采用fp16精度，无需量化。HuggingFace生态（Transformers、PEFT、TRL、Accelerate）在ROCm上无缝运行，无需修改代码即可直接替代CUDA。模型已上传至HuggingFace Hub并提供在线Demo。
Showing the latest 36 of 76 stories.