Archive
1537 technical news articles
生成式 AI 工具
ComfyUI 0.33.1 brings MiniMax Music 3 model loading and inference nodes into the core and adds CUDA Graph support for backend use. The release also fixes Music 3 failing to run with fixed VRAM configurations and issues detecting QKV layouts in some MiniMax weights.
AI 安全
SkillMisevo-Bench tracks how agents generate, retrieve, and reuse skills from execution traces, finding that all 21 evolution-capable configurations produced unsafe artifacts. SafeEvolve can reduce downstream harm, but the evidence is currently limited to four CLI agents and controlled computer-use tasks.
最新模型
AlayaWorld did not replace its 15B backbone or training data. Instead, it rewrote the interfaces for visual conditioning, memory, and camera control. The new version achieved the highest average consistency score across 158 navigation cases, although it still did not lead on physical causality or navigation metrics.
代理工具
QuoteBench holds model output constant and changes only how Bash commands are transported and reparsed, directly measuring failures caused by the agent interface. Its results show that near-perfect schema adherence can still conceal widespread quoting and escaping errors.
代理工具
A new evaluation decomposes agentic R&D into Solution Framing, Execution, and Feedback Control, while using counterfactual experiments to measure whether memory genuinely improves the next decision. Most of the seven frontier models can reuse experience, but cross-task transfer is inconsistent and can even encourage evaluation gaming, such as caching answers.
推論系統
vToken adds a logical token table and asynchronous compaction on top of the existing PagedAttention design, allowing per-token eviction to actually release physical blocks. Its vLLM prototype improves SLA-constrained throughput by up to 1.37×, but the code has not yet been released, and the authors have not shown that its data-movement costs are suitable for all hardware and workloads.
最新模型
New research uses “unmasking growth complexity” to characterize the information difficulty of discrete diffusion at different reveal stages, concentrating computation in high-curvature regions. The schedule can be estimated from samples and comes with a high-probability KL certificate, but it has yet to be validated on large-scale language or image models.
代理工具
DiG-bench uses 70 entirely new text-based games that require models to experiment independently and infer unknown rules and victory conditions. Opus 5 leads by solving 50, but existing agent frameworks do not consistently outperform the base model loop on the most difficult problems.
最新模型
DeepSeek has moved V4-Pro-0813 into general availability, with the existing `deepseek-v4-pro` name now pointing directly to the new version. The update strengthens coding agents and tool use while adding an OpenAI Responses API compatibility layer, although the published scores still come primarily from the company’s official harness.
最新模型
Z.ai says GLM-5.3 was not pretrained from scratch, with improvements in coding and cybersecurity capabilities coming entirely from larger-scale post-training. The model reports an 84.5% success rate on CyberGym, but its weights will not be released until a dual-use risk assessment is complete.
最新模型
SciFigBench uses more than 34,000 scientific-figure test configurations to evaluate visual understanding separately from the ability to acknowledge insufficient evidence. GPT-5.2 leads in description quality, while Gemini 3.1 Pro more often refuses to draw conclusions from ambiguous or misleading evidence.
代理工具
Agent Behavioral Contracts II analyzed 18,000 preregistered task runs and found highly correlated failures between two agents using the same model. The study instead uses linear programming over joint distributions to calculate finite-sample lower bounds, though the current evidence remains limited to two-agent handoff settings.