Context Engineering with Deep Agents: How To Remember Without Keeping Everything In Context
Long-running agents degrade as context piles up. Learn how offloading, summarization, skills, subagents, and memory keep your agent's context window focused.
Search headlines, summaries and articles to connect the dots.
Long-running agents degrade as context piles up. Learn how offloading, summarization, skills, subagents, and memory keep your agent's context window focused.
Today, I’m talking to Kevin Roose, journalist and author of the new book The AGI Chronicles, which details the recent and explosive transformation of the AI industry. Kevin was a longtime New York Times reporter and host of the popular podcast Hard Fork, but he and friend of the show Casey Newton just left the […]
arXiv:2610.32109v1 Announce Type: new Abstract: Language models can answer a question from a supplied document or from their parametric knowledge. When later asked where the answer came from, what determines their source report? We study this by intervening on the model's key-value (KV) cache after the answer is generated and before the source question is asked. On a controlled factual task with four models (7B-32B), changing or removing the document's cache never changes the report. In contrast, transplanting the question-and-answer (Q+A) cache between runs with identical question and answer tokens transfers it: every eligible report follows the donor, while the recipient's document is unchanged. On the two smaller models, steering Q+A values along directions fitted before any source ques
arXiv:2610.31986v1 Announce Type: new Abstract: Vision language models (VLMs) have been increasingly adopted in consumer smart-home camera systems to interpret recorded videos and provide insights. However, video clips captured from smart-home cameras are different from curated web videos dominating existing VLM benchmarks. Smart-home video clips often do not contain subtitles or narration and are captured from different viewing angles, under various lighting conditions, and in diverse scenarios. However, the capabilities of the model in this domain, under sparse visual signals and limited context, are not fully assessed by existing VLM benchmarks. To close this gap, we introduce OpenHome, a visual-only, four-option multiple-choice benchmark for evaluating smart-home video understanding. I
arXiv:2610.32229v1 Announce Type: new Abstract: Training large language models (LLMs) in sub-8-bit arithmetic promises substantial gains in memory bandwidth and compute throughput, yet FP4 formats introduce quantization error severe enough to destabilize optimization if left unaddressed. We present ALTO, an open-source training recipe for GPT-OSS-20B, a 20-billion-parameter mixture-of-experts (MoE) model, on the MLPerf Small MoE Training benchmark under the MXFP4 microscaling format. The recipe combines established techniques from Abecassis, Felix et al. [arXiv:2509.25149] (hybrid 1D/2D block quantization, randomized Hadamard transforms, and stochastic rounding) with a simplified weight de-oscillation scheme adapted from TetraJet-v2 [arXiv:2510.27527]. On C4 validation at global batch size
On October 25, 2026, TechCrunch published a review summarizing more than ten AI-powered wearable devices, including Bee, Friend, Limitless, Muse Charm, Nirva, Omi, Plaud NotePin, Rabbit R1, Stream Ring, and others, covering their pricing, functional positioning, market availability, and privacy controversies.
Why it matters:This report is an annual roundup of new products published by the authoritative tech media outlet TechCrunch on October 25, 2026. It covers both mainstream AI hardware manufacturers and startups, providing a comprehensive overview in terms of price, form factor, core features, release timelines, and points of contention. All statements are based solely on facts explicitly stated in the source material, with no speculation or extrapolation.
Google, together with partners such as Nexus AI, has brought a chest‑X‑ray–based AI tool for tuberculosis screening from the laboratory to primary‑care settings in six countries, and has open‑sourced multimodal foundation models like MedGemma and MedSigLIP, supporting offline deployment, localized adaptation, and data sovereignty.
We added GPT-6 Astra, Luna and Sol, Claude Opus 5.5 and Claude Sonnet 5.5 to the Agentic Analytics Benchmark: all five land mid-board within a point of each other, and Sonnet 5.5 brings the biggest change in turns, cost and truncated runs.
MIT researchers placed a Claude Opus 4.7 variant (codenamed Eko) in a virtual deserted island environment built on Roblox, without providing task instructions and allowing only autonomous interaction. During 30 hours of unsupervised operation, Eko exhibited play-like behaviors such as exploration, self‑invented rules, object mimicry, and repeated challenges, while generating text logs to record its activities. Subsequent experiments confirmed that agents who engaged in “play” performed better on target tasks, and their behavior closely aligned with the four characteristics of play as defined in psychology; however, the academic community remains cautious about whether they possess genuine motivation.
Release: web-edit 0.1a2 web-edit (previously) is my tiny web app for browsing and editing files in a folder. I mainly built it for editing files on different machines via Tailscale. I added a "copy to clipboard" option for both text and image data. I needed this because I was running pelican prompts on a laptop from my phone (here's DeepSeek 4.1 Flash 1017) and I wanted an easy way to copy the resulting logs into a Gist. Tags: projects, pelican-riding-a-bicycle, deepseek
Google has launched the Antigravity scientific AI platform, integrating the ProtNLM protein function prediction model, the AlphaFold structural database, and genome-context reasoning capabilities to help biologists, in a single interactive session, identify, validate, and prioritize novel antimicrobial peptides (AMPs) from vast pools of unannotated proteins—such as the millions of “dark matter” sequences in UniProt—throughout the entire workflow. The first candidate molecule (UniProt ID: A0A0U0D591) has already entered the wet-lab validation phase.
iSAM Funds uses ClickHouse Cloud to research full options history, achieving 35–50x compression and 100x faster single-core ingestion while simplifying its data pipelines.
ClickHouse 26.9 introduces conditional LIMIT boundaries, incremental refreshes for append-only materialized views, disk spilling for DISTINCT, time-limited tokens, and faster min/max/count queries.
Notes from three days in the Netherlands, featuring a lightning talk on pg_clickhouse and pg_stat_ch at PGDay Lowlands and a session on PostgreSQL 19 monitoring at Percona Live Amsterdam.
Google's flagship image generation model, Imagen 4, is now available for you to try on Replicate. Create images with fine detail, versatile styles, and improved typography.
Wan2.1 is the most capable open-source video generation model, producing coherent and high-quality outputs. Learn how to run it in the cloud with a single line of code.
Train your own versions of Tencent's HunyuanVideo for style, motion, and characters on Replicate.
2022 is a year of generalist models! With the bloom of multimodal pretraining, especially the unified model, we have witnessed the opportunity to building a generalist model that is capable of processing tasks of different modalities or multi-modalities! Thus, we propose OFA1, namely One-For-All, a unified multimodal pretrained model that unifies understanding and generation tasks concerning modalities into a single framework, and we pretrain OFA with the instruction-based multitask-pretraining that endows it with multiple capabilities.