今週のストーリー
埋め込み (Embeddings) × 開発者ツール

Cohere、最新埋め込みモデル群Embed 5を発表

Cohere、最新埋め込みモデル群Embed 5を発表

z = 3.0 5社横断 42本 2026-09-30（±1日窓） /ja/ai/story/embeddings-2026-09-30
✎ ストーリー本文

CohereがEmbed 5を発表し、同週は埋め込み・検索まわりの研究も重なった。

何が起きたか

Cohereが次世代の埋め込みモデル群「Embed 5」を発表した。同じ週にarXivでは、文脈内例のみで行う高密度検索や、長文脈埋め込みの選択的プーリングなどの論文が出ている。

なぜ重要か

埋め込みはRAGや検索の土台で、ベンダーの新世代モデルと検索精度・長文脈対応を巡る研究が並んだ形だ。NVIDIAのAIストレージ基盤の発表も同日にあるが、直接の関連は確認できていない。

次に何を見るか

Embed 5の性能・価格・対応言語の公開内容と、研究側の手法が実製品の検索に取り込まれるかを追いたい。

▲ 公式・報道
公式

Introducing Embed 5—A New Family of Frontier Embedding Models

Cohere Blog ・ 2026-09-30 ・ 📌

Cohere、企業検索向け埋め込みモデル「Embed 5」Pro/Fastを発表

公式

Expanding AI Storage Access with NVIDIA cuObject and the NVIDIA SCADA Server SDK

NVIDIA Developer Blog ・ 2026-09-30

NVIDIA、AI向け高速ストレージ接続を担うcuObjectとSCADAサーバーSDKを紹介

学術（arxiv ほか） 40本 ▾
学術

STEPQuant: When and Where Errors Matter in Delta-Rule Recurrent State Quantization

arXiv cs.CL (Computation and Language) ・ 2026-09-29

学術

LeapQuant: Efficient Linear Attention with Accurate Recurrent State Quantization

arXiv cs.LG (Machine Learning) ・ 2026-09-29

学術

Thinking Before Thinking: Scaling Agentic Inference Through Meta-Reasoning

arXiv cs.AI (Artificial Intelligence) ・ 2026-09-29

学術

Learning Meta-Skills for Agent Harness Design in Test-Time AI4AI

arXiv cs.CL (Computation and Language) ・ 2026-09-29

学術

Breaking the Uniformity Trap: Scaling Video Diffusion Model via SplitMoE

arXiv cs.AI (Artificial Intelligence) ・ 2026-09-29

学術

LongHarness Bench: Stress-Testing Language Model Harnesses for Long-Context Reasoning

arXiv cs.CL (Computation and Language) ・ 2026-09-29

学術

Achieving an $O(1/N)$ Optimality Gap in Average-Reward Weakly-Coupled MDPs

arXiv cs.LG (Machine Learning) ・ 2026-09-29

学術

WUSH-KV: KV Cache Quantization with Data-Adaptive Transforms

arXiv cs.LG (Machine Learning) ・ 2026-09-29

学術

ReCIRC: Rectified Conformal Risk Control

arXiv cs.LG (Machine Learning) ・ 2026-09-29

学術

From Routing Signals to Selective Review: Visual regrounding in MoE VLMs

arXiv cs.CL (Computation and Language) ・ 2026-09-29

学術

Effective Dense Retrieval using Only In-Context Examples

arXiv cs.CL (Computation and Language) ・ 2026-09-29

学術

Probe-Space Preconditioning for Fast and Stable Zero-Order Training

arXiv cs.LG (Machine Learning) ・ 2026-09-29

学術

Mira: Memory-Efficient MoE Inference Using Adaptive Caching and Predictive Expert Staging

arXiv cs.LG (Machine Learning) ・ 2026-09-29

学術

Probability is Not Enough: Exploring and Counting Divergent Tokens for Reasoning Uncertainty Quantification in LLMs

arXiv cs.AI (Artificial Intelligence) ・ 2026-09-29

学術

A foundation model for energy and radiation systems built on heterogeneous scientific interfaces

arXiv cs.LG (Machine Learning) ・ 2026-09-29

学術

UserProxyBench: Evaluating LLM User Simulators for Agent Benchmarks and Training

arXiv cs.AI (Artificial Intelligence) ・ 2026-09-29

学術

Dr. OPD: Learning What to Follow for Optimal On-Policy Distillation of Large Language Models

arXiv cs.CL (Computation and Language) ・ 2026-09-29

学術

Retrieval-Augmented Skill Optimization via Cross-Harness Adaptation

arXiv cs.AI (Artificial Intelligence) ・ 2026-09-29

学術

PE-EK-PINN: Physics Embedding with Evolving Kernel for Scalable Physics-Informed Neural Networks

arXiv cs.AI (Artificial Intelligence) ・ 2026-09-29

学術

Auditable Long-Term Memory: A Deterministic Retrieval Chain Measured at 479/475 of 500 on LongMemEval-S

arXiv cs.CL (Computation and Language) ・ 2026-09-29

学術

BITEM at the NTCIR-19 R2C2 Task: Predicting Confidence from Agentic RAG Pipeline Signals

arXiv cs.CL (Computation and Language) ・ 2026-09-29

学術

Which Attention Heads are like the Human Head? Not the Ones that Compute

arXiv cs.AI (Artificial Intelligence) ・ 2026-09-29

学術

$S^3$: Spectral Null-Space Swap Makes Reasoning Models Efficient

arXiv cs.CL (Computation and Language) ・ 2026-09-29

学術

Dagger: Decoupling-based Model Stealing Attack against Graph Neural Networks

arXiv cs.LG (Machine Learning) ・ 2026-09-29

学術

Kolmogorov-Arnold Classifier Systems as Universal Approximators

arXiv cs.LG (Machine Learning) ・ 2026-09-29

学術

Post-Anomaly Detection Inference for Deep SVDD

arXiv cs.LG (Machine Learning) ・ 2026-09-29

学術

Learning What to Remember: Long-horizon Counterfactual Memory Optimization

arXiv cs.CL (Computation and Language) ・ 2026-09-29

学術

RLX: A Unified Multi-Backend Tensor Compiler and Distributed Runtime in Rust

arXiv cs.AI (Artificial Intelligence) ・ 2026-09-29

学術

Overcoming Scaling Limits in On-Policy Self-Distillation for LLM Reasoning

arXiv cs.CL (Computation and Language) ・ 2026-09-29

学術

Beyond Interaction Capacity: Estimator Scaling with Recursive Models for CTR Prediction

arXiv cs.AI (Artificial Intelligence) ・ 2026-09-29

学術

Guide, Then Let Go: Gap-Adaptive Teacher Scheduling for Sparse-Reward Agentic RL

arXiv cs.AI (Artificial Intelligence) ・ 2026-09-29

学術

How Many Labels Does a Language Need? Annotation Budgets and Cross-Lingual Pooling for African-Language Text Classification

arXiv cs.CL (Computation and Language) ・ 2026-09-29

学術

Learning Beyond What You Sample: Off-Policy-Aware Cross-Model Trajectory Exchange for RLVR

arXiv cs.CL (Computation and Language) ・ 2026-09-29

学術

One Threshold Does Not Fit All Languages: Language-Conditional Deferral for Reliable and Efficient Low-Resource Text Classification

arXiv cs.CL (Computation and Language) ・ 2026-09-29

学術

Storage Is Not Strategy: State-Conditioned Support Control for LLM Unlearning

arXiv cs.CL (Computation and Language) ・ 2026-09-29

学術

Can Vision-Language Models Stay Helpful When Facing Implicit Risks? Intent-Privilege OPSD for Efficient Safety-Helpfulness Alignment

arXiv cs.CL (Computation and Language) ・ 2026-09-29

学術

The Geometry of Inference in Transformer Residual Streams

arXiv cs.CL (Computation and Language) ・ 2026-09-29

学術

Selecting What Matters: Semantic Compression-Guided Selective Pooling for Long-Context Embeddings

arXiv cs.CL (Computation and Language) ・ 2026-09-29

学術

Billiger.de Products: A Bilingual Entity Matching Benchmark

arXiv cs.CL (Computation and Language) ・ 2026-09-29

学術

Corpus-Guided Dual-Path Propagation for Graph Retrieval-Augmented Generation

arXiv cs.CL (Computation and Language) ・ 2026-09-29

← ストーリー アーカイブ