AI Intelligence Archive

TL;DR DeepInfer is the inference engine DeepSeek built and runs behind its own service,...

I've been building FableCut for the past few months, and one thing I've wanted to explore is what...

n8n has published a guest contribution with Cognee that outlines a practical approach to giving AI...

While language models have become dramatically more capable, retrieval still largely relies on lexical matching or similarity in fixed embedding spaces. This creates a capability asymmetry: what the retriever fails to find, even the strong…

As autonomous research systems become more capable, benchmarks must both evaluate their performance and support meaningful research progress. We introduce Auto-Research Arena, a multi-target benchmark for evaluating autonomous research sys…

Evolution strategies (ES) update a model by sampling a population of perturbed copies, scoring each one, and moving in the reward-weighted direction of the perturbations. This update is usually presented as an estimate of a smoothed gradie…

Many everyday tasks require controlling contact force as well as motion. On a humanoid, this force must often be coordinated across both hands and the whole body, and cannot be specified by kinematic trajectories alone. Therefore, we prese…

Learning algorithms are usually analyzed under the assumption that informative gradients are continuously available. But this is not always true---learning signals in biological neural networks, for example, may be sparse, local, and only …

Large language models (LLMs) show promise in specialty medicine, yet existing evaluations largely focus on isolated clinical encounters rather than tracking how clinical responsibilities evolve across a longitudinal care pathway. Here we i…

Expanding multilingual ASR systems to new languages is challenging when adaptation must preserve previously supported languages without replaying source-language training data. We study this problem in connector-based Speech LLMs, where a …

As foundation models evolve from information generation to digital work execution, agents are increasingly capable of completing complex tasks across tools, applications, and workflows. However, existing research lacks a unified perspectiv…

Human oversight is a pillar of AI safety. It is often exercised by a human in the loop (HITL) who approves critical actions taken by an AI agent. AI agents' increasing ability to bypass guardrails, though, raises the question of how to enf…

In federated learning (FL), secure aggregation enables a server to compute the sum of local model updates of $N$ clients, without revealing any information on the individual model updates to the server, beyond their sum. Different clients …

Frontier multimodal large language models (MLLMs) can control robots without robot-specific training, but how their capabilities compare and interact with post-trained vision-language-action (VLA) and world action model (WAM) policies acro…

A vulnerability costs more to fix the later it is discovered. As coding agents generate code faster than developers can review it, software needs a first line of defense screening each new function in real time. LLM agents have recently be…

Selection bias in large language models (LLMs) can persist despite instructions to ignore presentation order or prompts that elicit reasoning. Recent efforts to reproduce Jev-style System One behavior have produced a growing family of mode…

From Figma integrations to AI coding agents, here is how to choose the right tool for turning an...

A robot policy needs to understand what a task is asking for in a scene, and how to move the robot to accomplish that task. To achieve these two skills, we introduce Zero-real Agent Policy (ZAP), a recipe that teaches a small off-the-shelf…

Feed-forward 3D reconstruction models such as VGGT regress cameras and geometry in a single forward pass, but they report no uncertainty for the camera poses they predict: an incorrect relative pose looks no different from a correct one. W…

Diffusion large language models (dLLMs) can generate multiple tokens in parallel, but often require many decoding rounds to complete a response. Efficient generation depends not only on which tokens are easy to predict, but also on how com…

Many academic disciplines (in particular, mathematics and the theoretical sciences) face difficult questions about how to adapt their practices to the rise of highly capable artificial intelligence (AI). Before AI, a well-written research …

Sparse outcome rewards make it difficult for long-horizon language-model agents to assign credit to earlier actions. Learned critics provide value estimates for intermediate states, but these estimates can be inaccurate, while estimating i…

Deep research, or any use of large language models (LLMs) to read a large document collection and produce insights, requires synthesizing information from different sources. In this work, we find that LLMs produce long-form research report…

Preference optimization can learn spurious correlations, and augmenting the preference data with ties (equally preferred responses) can reduce spurious reliance. However, ties differ in their corrective effect, making their selection for a…

Last week I wrote about never having written a line of code before June. This is the next instalment,...

Four developer stories from October 26, 2026: what Anthropic's public S-1 reports, how OpenAI and Anthropic cut LLM prices, the vLLM flag that fixes open-model tool calls

Every engineer running autonomous agent loops knows the in-context reflection drill. The agent runs a...

I built a calorie tracker. It's called ForkCount. It's free. There's no card, no trial, no "start...

An Open Letter to Harun Harun, I almost never write articles addressed directly to a single...

We run several coding agents at once: Claude Code in two terminals, Codex in its app, sometimes...

From one prompt to a signed, open-source app running on my POCO M8 in a handful of iterations. Here's...

JetBrains open-sourced Thinkrail, an Apache-2.0 agentic IDE where every agent works in its own git worktree, the project's truth lives in a graph of SPEC.md files the agent maintains itself, and there are no accounts. Repo walkthrough: the…

ChatGPT may retrieve many sources for an answer, but cite only a small fraction of them. That...

On Oct 22, I read the word "Merged" twice. It was my first contribution to Forem (the platform that...

Meta has delayed the effective date for updated Commercial Terms and Self-Serve Ad Terms that contain...

One prompt with the /anatomy skill gave me an interactive isometric bench for an API request. Here's what each part means, what I changed after the first version, and how to remix it.

Browsers save files to Downloads without user interaction, while AI agents automatically follow instruction files. These two behaviors, neither a bug

Scheduling a coding agent takes a few minutes. A launchd plist, a cron line or a desktop scheduled...

There is a letter on my kitchen counter I have not opened in nineteen days. I know roughly what it...

Digital marketing involves a surprising amount of repetitive work. Before launching a campaign, you...