#Image generation

Huawei's Kirin 9050 series new Malang GPU and AI ISP intelligence disclosed: implements second-generation hardware ray tracing, realizing 200 million pixel fusion photography for the first time.

The Huawei Kirin 9050 series is equipped with the fourth-generation GPU micro-architecture, offering a 40% performance boost and a 45% reduction in power consumption. The second-generation hardware's ray tracing performance has improved by 78%, supporting BVH, FERT, and AO hardware acceleration units. The new-generation AI ISP integrates a pixel-specific NPU, achieving 200 million pixel fusion photography for the first time, supporting 4K60 video processing, 18EV high dynamic range video, and the full-chain BT.2020 wide color gamut.

Deploying OpenFold3 on Amazon SageMaker JumpStart for Predicting Biological Molecule Structures

OpenFold3-preview can be deployed on Amazon SageMaker JumpStart, an open-source biomolecular structure prediction model developed by the OpenFold Consortium and AlQuraishi Lab at Columbia University, supporting 3D structure prediction for proteins, DNA, RNA, small molecule ligands, and their complexes.

Using NVIDIA NeMo Switchyard on Amazon Bedrock to Route Model Requests for Coding Agents like Claude Code and OpenAI Codex CLI

Developers using coding agents like Claude Code and OpenAI Codex CLI need to be billed per task rather than per single call, so model routing should be based on the cost per task. NVIDIA NeMo Switchyard is an open-source library that helps AI agents select models to process each request, supporting Stage and Random routing strategies.

Run GPU-accelerated black hole interferometry using NVIDIA cuPhoton

Using NVIDIA cuPhoton, black hole interference measurements can be achieved, with the observation ring of M87* being only 42 microarcseconds. The black hole itself does not emit light; telescopes capture radio emissions from the hot gas outside the event horizon. The strong gravitational bending of light by the black hole forms a bright ring. NVIDIA cuPhoton accelerates related calculations through GPU.

DeepSeek releases the full suite application API

DeepSeek has opened up the application API, allowing developers to call deep thinking search, file input, and other dialogue functions through parameters. Voice and related underlying capabilities are also opened as independent interfaces, supporting separate or combined calls to achieve 1:1 replication of the DeepSeek App. The V4.1-Flash model (deepseek-flash) supports online search, deep thinking search, file input, reading links, and other capabilities, and charges fees for related tool calls.

Gentle Monster and Warby Parker will launch Android XR smart glasses next month equipped with Qualcomm AR1 chips and connected to Google Gemini.

ArtCraft company open-sources Photoshop alternative PhotoCraft and plans to refactor the internet

ArtCraft company open-sourced an AI-built, Rust-based Photoshop alternative PhotoCraft, with a 60% similarity in functionality to Photoshop (GitHub data shows 46% professional preparation), and plans to open-source a clone of Adobe Acrobat's PdfCraft.

Shanghai Hisilicon has launched the CS6286 image sensor, extending the endurance of the vehicle guardian system to over 7 days.

RAG reordering: How to use a reordering component to fix poor retrieval results

When RAG applications exist correct information in documents but provide wrong answers, it is usually due to errors in the retrieval stage. Vector search can find relevant passages on the topic, but it often fails due to approximate matching, such as incorrectly calling the bucket limit of the Enterprise plan when asking about the Growth plan.

DeepSeek-V4.1-Flash achieved a 58-fold throughput improvement on NVIDIA Vera Rubin NVL72.

DeepSeek-V4.1-Flash achieves a service capability of 1.7M tokens/s per GPU in a single NVIDIA Vera Rubin NVL72 rack, reaching up to 387 tokens/s per user, an increase of 58 times compared to GB300 NVL72.

Gemini and Z3 jointly capture bugs missed in Go code fuzz testing

Gemini 4 Argon translates Go code panic paths into Z3 constraints, paired with Z3 to check trigger conditions, and paired with go test to replay trigger bytes, locating 3 embedded parser bugs and discovering 3 skill-lint potential bugs within 1 minute. This process is more efficient than go test -fuzz, capable of finding boundary check errors missed by fuzz testing.

Phota Labs converts everyday photos into professional-level images.

The Phota Labs app uses AI to transform everyday photos into professional-grade images while preserving the authenticity of the photos. The app focuses on fixing issues with the photo itself, rather than enhancing already poor photos. Users can choose different photography styles for AI conversion and use personalized technology to learn the user's identity to maintain the authenticity of the photos. The app embeds invisible watermarks to identify AI-edited content, and it is currently launched in a free mode to build a user base.

Helion's Internal Compile-Time Heuristic Algorithm Optimizes Kernel Performance

Helion analyzes kernel IR and data flow through compile-time heuristic algorithms, selecting fast settings without slow automatic tuning, classifying kernels as matrix multiplication, reduction, or point-wise operations, and choosing block size, warps, and phases to achieve near-automatic tuning speed.

Amazon Bedrock Managed Knowledge Base implements fine-grained access control

Amazon Bedrock Managed Knowledge Base achieves fine-grained access control for enterprise AI agents by injecting user context into the AgentCore Gateway tool layer and verifying JWT, combined with the ACL-aware functionality of the knowledge base.

Galaxy Smart Learning's Little Deer Love Learning introduces a dual-core technology architecture to support scalable personalized instruction.

Galaxy Smart Learning has launched an AI education product, Xiao Lu Ai Xue, which adopts a dual-core architecture of the Doubao large model and a self-researched intelligent learning engine. It achieves learning situation diagnosis by analyzing students' answer records and develops functions such as learning situation diagnosis, ability assessment, learning difficulty matching, and task planning. The student-side intelligent agent recommends learning tasks and explains the basis, while the teacher-side intelligent agent summarizes student performance for personalized guidance reference. The product has served over 500 schools, covering 24 provinces, and supports teachers in adjusting teaching strategies through a human-machine co-education model.

EUROPA Union trains a 1T parameter MoE model using PyTorch and TorchTitan on the JUPITER supercomputer

The EUROPA Union successfully trained a MoE model with 951.9B parameters on the JUPITER supercomputer using PyTorch and TorchTitan, increasing single GPU throughput from 90 to approximately 1,300 tokens/s with 384 NVIDIA GH200 GPUs.

Apex Compute releases ANR Cross-Platform AI Rendering Technology test version, challenging NVIDIA DLSS

Apex Compute has released the ANR cross-platform AI rendering technology test version, which includes AI frame generation, super-resolution, and ray reconstruction features, compatible with NVIDIA, AMD, Intel graphics cards, and Steam Deck. Currently, only the AI frame generation feature is open for testing.

The core technical team of Tongxin Intelligent has started a business on AGI architecture chip, obtaining hundreds of millions of investment from Tsinghua System and other funds.

Tongxin Intelligence has completed tens of millions of yuan in seed and angel funding within half a year, with an average chip industry experience of over 20 years for the founding team. The company designs native chips supporting vector, tensor computation, and physical intelligence based on the CUV cognitive architecture proposed by Professor Zhu Songchun. The first product is positioned as a high-value inference chip, targeting Nvidia's Thor series on the edge and B300 series on the cloud, with FPGA prototype verification planned for early December. Funds will be used for core architecture design and R&D before chip fabrication.

How to Benchmark LLM with OpenRouter

Running the same tasks on OpenRouter, GPT-6.1 Sol passed 180/180 tests in 60 synthetic tasks, DeepSeek had the lowest cost passing 157/180 times, but the correlation with the public ranking is low (Spearman correlation coefficient 0.109). Need to check the number of task repetitions, the number of answers, and billing verification.

Gating Mechanisms of State Space Models (SSMs): Memory and Contextual Learning Research

State Space Models (SSMs) lag behind Transformers in contextual learning and precise retrieval tasks, as their gating mechanism leads them to first learn memory solutions, delaying or preventing the convergence of correct contextual learning solutions, but the gating mechanism is beneficial for improving long sequence generalization.

Hugging Face blog: Token Pooling-aware training for multi-vector embedding models.

Hugging Face's blog introduces the Token Pooling perception training method for multi-vector embedding models, which compresses storage by merging similar tokens through HierarchicalTokenPooling.

Xbox version of 'GTA 6' pre-load files show hints of AMD FSR 4.1, launch may not support

Data miners found libraries named amd_fidelityfx_loader_xbox.dll and amd_fidelityfx_upscaler_xbox.dll in the pre-load files of Xbox version 'GTA 6', consistent with the terms of the Microsoft FSR 4.1 Xbox GDK sample project, suggesting the game may adopt the latest upscaling technology; however, developer Rockstar has not confirmed, and this technology may be added through updates after release.