代理工具

Agent、MCP、coding agent、工作流自動化與評測。

StepGuard Reviews Agent Actions Before Tool Execution, but Adds About 2.53 Seconds of Inference per Task

A team including the Shanghai AI Laboratory has released StepGuard, a 4B-parameter model that uses paired safe/dangerous trajectories and Balance-GRPO to calibrate overblocking and missed detections. It substantially reduces attack success rates in two agent environments, but its training corpus and data generator have not yet been released, and it still misses some policy violations.

2026-08-27

推論系統

AgentX 1.0 Replays Million-Token Inference Workloads from 393 Real-World Agent Traces

SemiAnalysis has released AgentX, featuring long-context, multi-turn, and concurrent subagent workloads so inference systems are no longer compared solely using fixed-length prompts. The dataset preserves request timing and KV-cache reuse patterns, but its sources are concentrated on Claude Code, and it measures serving-system efficiency rather than programming-task correctness.

2026-08-26

代理工具

Perplexity Moves Its Full Agent Stack to DGX Spark, Sending Only Difficult Steps to the Cloud with User Approval

Portable Computer runs a 27B model, planner, tool routing, persistent task queue, and search index locally, while permission prompts control data transfers. The initial release supports only Linux and DGX Spark, and its official results come from an unpublished internal workload that should not be treated as independent validation.

2026-08-26

代理工具

DeepRepoQA Uses Monte Carlo Tree Search to Trace Cross-File Code Relationships, but Evaluation Still Relies on LLM Judges

Rather than having coding agents search along a single path, DeepRepoQA uses MCTS to compare multiple repository exploration branches and synthesizes answers from evidence with line-number citations. Its authors report gains of 4 to 7 percentage points on SWE-QA, but data contamination, inference costs, and limited cross-language testing remain to be validated.

2026-08-26

最新模型

Qwen3.8-Flash-Next Combines Linear Attention with Block-Sparse Retrieval, Previewing the Qwen4 Architecture

Alibaba has released the Qwen3.8-Flash-Next weights. The model uses Gated DeltaNet to compress most historical state, while Qwen Sparse Attention retrieves information from the full context. It natively supports 262K tokens, but the official million-token performance and agentic evaluations have yet to be independently reproduced.

2026-08-26

AI 安全

Root Access Can Abuse the Pixel Camera’s Hardware Key to Sign Fake Images, and C2PA Verification Will Still Pass

Security researcher David Buchanan used Keystork to impersonate the camera app and ask Android StrongBox to generate valid Content Credentials for arbitrary AI-generated images. The attack does not require exporting the private key; it exposes how a compromised signing endpoint can abuse a genuine key, rather than a flaw in C2PA’s signature algorithms.

2026-08-26

最新模型

Thomson Reuters Continues Training Qwen on 200B Tokens, Releases 35B Legal Model for Noncommercial Research

Thomson-1.0-Small extends Qwen3.6-35B-A3B through mid-training on proprietary documents, synthetic rewrites, and general-capability replay data. Official evaluations show improved legal-agent capabilities, though the model trails the base model on some general and coding tasks.

2026-08-26

最新模型

Granite Speech 5.0 移除 LLM 骨幹，以 470M CTC 模型換取逾 12,600 RTFx 批次轉錄

IBM 的兩款英語語音辨識模型把輸出速率降至每秒12.5個token，並以三段時間下採樣減少運算。官方在單張H200測得每秒處理逾3.5小時音訊，但這是大批次吞吐量，不等同互動串流延遲。

2026-08-26

最新模型

QAH 讓 60B MXFP4 模型在 9 項評測中有 7 項超越同架構 BF16 版本

Multiverse Computing 改由原始 120B 模型直接蒸餾經結構壓縮及量化的 60B 學生模型，避免把中間恢復模型的誤差當成學習上限。結果顯示 4-bit 模型可超越同架構 BF16 檢查點，但未全面追回原始教師的能力。

2026-08-26

代理工具

Claude Chat and Cowork Merge Persistent Memory, but the Two Interfaces Cannot Be Disabled Separately

Anthropic is giving chat and its task-execution interface, Cowork, a shared set of editable memories and will now update them during conversations. Sensitive topics are excluded by default, but memory is enabled by default on general consumer plans, and users cannot prevent only one interface from accessing it.

2026-08-26

最新模型

Tencent Open-Sources WeMM-Embedding: 2B Model Outperforms Some 8B Baselines in Multimodal Retrieval Benchmarks

WeMM-Embedding processes text, images, video, visual documents, and interleaved inputs in a shared vector space and is available in 2B, 4B, and 9B sizes. Official results show it outperforming several public baselines, but the model does not support audio, and some product-level results are backed only by Tencent’s internal A/B tests.

2026-08-26

訓練資料

LAION-BVD Releases a 10-Million-Hour Video Corpus, but Media Files Are Limited to Approved Noncommercial Research

LAION-BVD compiled 1.3 billion video URLs from Common Crawl and downloaded approximately 80 million videos totaling 10 million hours. The data pipeline and URL lists are public, but access to the main media subset is restricted, and researchers remain responsible for compliance.

2026-08-26