硬體

AI 晶片、GPU、資料中心、邊緣裝置與硬體安全。

AMD Helios Brings 72 MI455X GPUs Into a Single Memory Domain, Delivering 31 TB of HBM4 per Rack

AMD has officially unveiled the CDNA 5-based Instinct MI455X and the 72-accelerator Helios rack. The new platform challenges NVIDIA NVL72 with open interconnects and greater HBM capacity, although the performance figures disclosed so far are still primarily theoretical peaks.

2026-07-25

最新模型

Faster IndexTTS-2 Ports Its Autoregressive Speech Pipeline to TensorRT, Delivering Up to 3.6× End-to-End Speedup

A new study restructures IndexTTS-2’s GPT, Diffusion Transformer, and vocoder using TensorRT and TensorRT-LLM, while adding streaming and batched inference. Experiments on the English and Chinese Seed-TTS benchmarks show up to a 5× speedup for GPT, but the currently available public materials remain insufficient to directly reproduce the complete optimization pipeline.

2026-07-25

代理工具

NemoClaw 0.0.93 Improves Local Agent Installation on DGX, Makes Download Credentials Ephemeral, and Adds Resume Support

NVIDIA has updated NemoClaw’s installation process for DGX Station and DGX Spark to prevent existing vLLM workloads from being stopped accidentally and to improve credential and rate-limit handling for large model downloads. The release remains in alpha, and the newly supported DGX OS configurations have not yet completed end-to-end qualification.

2026-07-25

最新模型

Prime Intellect Converges RLVR, LoRA, and Deployment for Nemotron 3 Nano into a Managed Training Workflow

NVIDIA has released a reproducible managed post-training workflow that lets developers use the Prime CLI to run 100 steps of RLVR on a 30B MoE model, then download or deploy the resulting LoRA. In the demonstration, performance rose from 21.9% to 90.6% across 32 math problems, though the sample was small and both the tasks and rewards were tightly controlled.

2026-07-25

最新模型

CompactifAI Compresses Llama 3.3 70B to About 65 GiB, Runs It on a Single Xeon 6 with vLLM CPU

Multiverse Computing has unveiled a compressed and healing-trained version of Llama 3.3 70B that can be served on Xeon 6 using vLLM CPU and Intel AMX. Official tests show nearly double the throughput, but output remains below 4 tokens per second, and neither the checkpoint nor the raw results have been released.

2026-07-24

推論系統

MoAI Showcases GLM-5.1 on 32 MI300X GPUs, Integrating Heterogeneous GPU Routing and Disaggregated Inference into a Single Service

Moreh demonstrated a distributed inference service running GLM-5.1 across four nodes equipped with 32 AMD MI300X GPUs, with live displays of TTFT, TPOT, and GPU utilization. MoAI further claims it can split prefill and decode across GPUs from different vendors, but the latest demonstration did not disclose complete throughput data or a reproducible configuration.

2026-07-24

最新模型

Microsoft Specializes MAI-Code Through In-Product Reinforcement Learning, Extending the Same Checkpoint to an Excel Agent

Microsoft has unveiled a “hill-climbing” post-training process that feeds product tools, graders, and real-world usage signals back into model iteration. After further training in an Excel environment, MAI-Code-1-Flash can be served on older GPUs, but the company has not disclosed its parameter count, evaluation sets, or raw experimental data.

2026-07-24

代理工具

Prime Intellect Unifies 365,000 Agent Tasks, Giving Evaluation and Reinforcement Learning a Shared Execution Interface

Prime Intellect has consolidated software engineering, terminal operation, and web search datasets into 23 tasksets accessible through a unified API for agent evaluation and reinforcement learning. The platform also delays injecting tests and grading materials to reduce reward hacking, but grading still shares a sandbox with the agent, so isolation remains incomplete.

2026-07-24

推論系統

PyTorch Helion Adds TPU Backend, Bridging Triton and Pallas with the Same High-Level Kernels

PyTorch and Google have built a TPU backend for Helion, allowing PyTorch-style kernels to compile to Pallas and reducing the need to write separate low-level TPU kernels. The team measured up to 838 TFLOPs for Flash Attention on TPU v7, though the feature remains under active development.

2026-07-24

AI 安全

ResearchArena Tests Deliberate Sabotage by AI R&D Agents; Data Poisoning Is Harder for Monitors to Detect Than Code Backdoors

ResearchArena turns post-training, CUDA kernel optimization, and inference server optimization into long-horizon evaluations with hidden attack objectives. Across 384 experiments, allowing monitoring agents to execute the resulting artifacts helped uncover backdoors, but sabotage embedded in training data still frequently went undetected.

2026-07-23

AI 基礎設施

GigaToken Accelerates BPE With Rust, SIMD, and Direct File I/O, Reaching 24.53 GB/s in a GPT-2 Tokenization Benchmark

The open-source GigaToken 0.9.0 project reoptimizes pre-tokenization, BPE merging, and the data input path, claiming a 989× speedup over Hugging Face Tokenizers on a dual-socket, 144-core AMD EPYC server. Its top result depends on the native API, an 11.9 GB corpus batch, and highly parallel hardware, and cannot be directly extrapolated to short online requests.

2026-07-23

最新模型

Solar Open 2 Supports a Million-Token Context with Hybrid Linear Attention, Activating Just 15B of Its 250B MoE Parameters per Token

South Korea’s Upstage has released Solar Open 2 for long-horizon agents, combining a 250-billion-parameter MoE, hybrid attention, and a 1-million-token context window in a single model. The company says the quantized model can be deployed on two NVIDIA H200 GPUs, though current evidence for its performance and long-context stability comes primarily from the team’s own testing.

2026-07-23