最新模型

新模型、開放權重、多模態、語音、影像與模型架構。

MCP 2026-07-28 Removes Connection-Phase Sessions as Tool Servers Move to Stateless HTTP

The new MCP revision removes the initialization handshake and `Mcp-Session-Id`, allowing any server instance to handle an individual request. This is a breaking protocol change; deployers should retain version negotiation and fallback to the previous version until the final release and major SDKs are fully aligned.

2026-07-28

代理工具

NVIDIA Open-Sources NOOA, Unifying Agent State, Tools, and Persistent Memory in a Python Class

NOOA represents an agent as a single Python class, replacing fragmented prompts and tool schemas with methods, fields, docstrings, and type annotations. NVIDIA reports 82.2% on SWE-bench Verified with roughly half the token usage, but the framework remains a research preview and requires a separate sandbox when executing model-generated code.

2026-07-27

最新模型

Kimi K3 Full Weights Officially Released, Taking the 2.8T MoE from API Access to Self-Deployment

Moonshot AI released Kimi K3’s full weights, model card, and technical report on July 27, enabling external teams to deploy and audit the nearly three-trillion-parameter sparse model. However, requirements including at least 64 accelerators, custom code, and dependence on reasoning history create clear barriers.

2026-07-27

GitHub Repo

Nanbeige4.2-3B Compresses an Agent Model With a Looped Transformer, Supporting a 256K Context With 3B Non-Embedding Parameters

Nanbeige Lab has released Nanbeige4.2-3B, pretrained from scratch on 28 trillion tokens. It repeatedly applies the same set of Transformer layers to increase computational depth without adding parameters. In official agent benchmarks, it generally outperforms the larger Qwen3.5-9B, but the results depend heavily on Nanbeige's own execution framework and still require independent reproduction.

2026-07-27

最新模型

SGLang 0.5.16 Dynamically Adjusts Speculative Verification Length Based on Confidence and Overhauls Caching for Hybrid Architectures

The new SGLang release introduces DSpark, which changes the verification window for each round based on the draft model’s confidence, and makes UnifiedRadixTree the default cache for sliding-window, Mamba, and sparse-attention models. The release also removes several quantization backends and changes the rollout transport format, so compatibility must be checked before upgrading existing services.

2026-07-27

最新模型

Cosmos-H-Dreams Advances Surgical World Models Toward Real-Time Closed-Loop Operation, Generating About 160 FPS on a Single RTX PRO 6000

NVIDIA has unveiled a generative simulator for surgical robotics built with few-step diffusion, self-forcing distillation, and a streaming KV cache. The system can receive robot motion data in real time and synthesize subsequent video, but visual realism remains no substitute for physical correctness and clinical validation.

2026-07-27

最新模型

LLaDA 2.2 Lets Diffusion Models Add and Delete Tokens Mid-Generation, Targeting Iteratively Correctable Agent Loops

inclusionAI has open-sourced the 100B-class LLaDA2.2-flash, using Levenshtein Editing to let parallel decoding insert, delete, and replace content. The model outperforms the company’s own autoregressive baseline in some agent benchmarks and throughput tests, but still trails in code repair and function-calling accuracy, while serving support remains incomplete.

2026-07-27

推論系統

vLLM 0.26 Expands Cross-Layer KV Caching and Hybrid Attention, Accelerates DeepSeek-V4 Inference Kernels

vLLM 0.26 allows each KV-cache group to use its own attention backend and adds object storage to its tiered cache. The release also focuses on optimizing DeepSeek-V4, ROCm, XPU, and speculative decoding, though its official microbenchmarks do not directly represent real-world serving throughput.

2026-07-27

代理工具

FluxBench Puts Chip Agents Through the Full RTL-to-GDS Flow, With Performance Gaps of Up to 86% on the Same Model Architecture

FluxBench uses standardized prompts, tools, and technology libraries to test agents across the full flow from RTL generation through synthesis, placement and routing, and ECO, rather than evaluating only one-shot code output. The authors report that different agent frameworks using the same base model can exhibit a performance gap of 86.27%, although the commercial EDA experiments are limited in scope and the paper was previously withdrawn and resubmitted.

2026-07-27

代理工具

Dynamic Forecasting Leaderboard Reveals the Advantage of Specialized Agents: Frontier General-Purpose Models Do Not Dominate ForecastBench

ForecastBench’s preliminary leaderboard, updated July 27, shows specialized Torchcast AI agents taking all three top spots, with the highest Brier Index reaching 65.6. The GPT-5.5 search agent and the median superforecaster score fall within a similar range, but differences in question counts, submission timing, and confidence intervals limit direct ranking comparisons.

2026-07-27

推論系統

Nativ Packages MLX-VLM, Local APIs, and Performance Telemetry into an Open-Source macOS Inference Workspace

Nativ v0.0.1 integrates model downloads, chat, monitoring, and OpenAI- and Anthropic-compatible endpoints into a native SwiftUI app, recently drawing attention in the local AI community. The initial release supports only Apple Silicon and macOS 26, and no cross-model performance or compatibility benchmarks have been published.

2026-07-27

推論系統

AMD Open-Sources Hyperloom to Automatically Rewrite and Validate GPU Inference Optimizations Through a Multi-Agent Closed Loop

AMD has released ROCm Hyperloom 1.0.0a1, enabling agents to handle everything from performance profiling to kernel and host-code modifications, then decide whether to retain changes based on correctness and speed tests. The initial release supports only three Instinct GPUs, and the claim of reducing optimization time “from weeks to hours” remains AMD’s own; adopters will need to reproduce the results with their own models.

2026-07-27