The technique trains small neural networks offline to route and rewrite queries, achieving up to 8x faster search and higher recall than centroid-based indexing on the Natural Questions dataset.
The update enables developers to build and deploy agents, workflows, and apps across cloud, VPC, or on-premises environments using the Mistral Large 4 model.
Analysis of 207,049 sessions reveals that scientists use Claude most for tasks where AI performs well, such as data analysis and writing, while underutilizing it for physical data collection and ideation.
Gradium's Phonon TTS, available in 31m and 40m sizes, tops the Hugging Face Open TTS Leaderboard's CPU streaming benchmark for time-to-first audio.
The model features a Sparse MoE architecture with 14B active parameters per token and outperforms DeepSeek-V4-Flash-Base on MMLU and coding benchmarks.
Enterprise users can deploy LlamaParse in their own AWS, Azure, Google Cloud, or Kubernetes environments, with flexible options for managing model inference.
The library uses a shadow interpreter to parallelize independent IO-bound calls in sequential code without requiring async syntax, reducing runtime from 148s to 24s in benchmarks.
The protocol uses a Teacher-Orchestrator-Student structure to generate evolving problems, with the Orchestrator increasing problem validity from 4.30 to 4.85 and approval rates from 38.14% to 87.14%.
The system identifies borderline cases to help users refine instructions, reducing effort and stress compared to manual labeling.
The preliminary study used hundreds of GPUs for microscopy and produced a first structural model 34 hours after purification, though the biological function of the particles remains unknown.
The managed service handles data validation, type conversion, and compaction, allowing users to pay only for copied data without provisioning infrastructure.
They unveiled a demo that performs drug design iterations, which typically take days, in about 1 hour.
Kakao has released the 3B-scale multimodal decision-making model Kanana-2-v-decide as open source.
The method identifies that FP4 errors stem from low-entropy tokens and uses trace-aligned training to reach 65.94% AIME-120 accuracy on R1-Qwen-14B.
The open-source engine reduces batch-8 decode latency to 24.2 ms by fusing layers into single NKI kernels to minimize HBM traffic and compiler overhead.
Competing statements in context cause 9-110x higher loss than neutral filler, while length alone explains only 2-8% of variance.
The new model features significant improvements over the Preview version and includes a detailed model card with training and architecture explanations.
The migration to Cloud Hypervisor-based microVMs achieves 500 ms sandbox acquisition and avoids 96.1% of recent Linux kernel CVEs.
The new PCIe card features 64 AI cores, 576 MB on-die SRAM, and 548 GB/s DRAM bandwidth for generative AI workloads.
The 65B parameter model achieves a TerminalBench 2.1 score of 80.52, outperforming larger competitors like Reflection Beam 501B and Meta Muse Spark Open 816B.
New quantization files improve accuracy by 10-15% on agentic benchmarks and introduce a new engram switching process for multiple precisions.
The 276B-A12B model achieves 20.0% on Terminal-Bench 4.0 and 24.0% on FrontierCode 2.0, running at 200 tokens per second.
The 2B parameter model achieves a 0.834 field-level F1 score, surpassing Claude Opus 5 by 1.8 points and GPT-5.6 Luna by 9.3 points.
Kino AI offers a free Electron-based video editor built on Remotion that allows AI agents and human teammates to collaborate in real-time on a shared timeline.
AI Briefing