All Stories

Zyphra releases ZUNA1.1 as an Apache 2.0 EEG foundation model

fal brings a generative media and CUDA optimization workshop to Bengaluru

Yann LeCun links neural network generalization to inductive bias and regularization

Maharshi announces a hands-on kernel development session in Bangalore

Gemini app adds reusable Nano Banana avatars for image generation

Higgsfield MCP chains GPT-5.6 Sol, Seedream and Seedance into a 3D cartoon studio

Sam Altman says ChatGPT's new voice model crossed a usability threshold

fal and Decart enable live webcam restyling with Lucy 2.5 Realtime

John Carmack says Claude caught a CUDA clamp needed for sqrt epsilon safety

Google DeepMind and Isomorphic Labs outline frontier AI bioresilience plan

Reddit report claims GPT-5.6 solved all six IMO 2026 problems

Air release adds coding-agent imports and Java support

Simon Willison ports Grok Build's Mermaid renderer to WebAssembly

xAI open-sources Grok Build's Rust coding-agent repo

Prof. Tom Yeh walks through a Graph Convolutional Network by hand

Sakana AI adds NVIDIA Nemotron models to Fugu orchestration

Google Research explains diffusion model creativity via score smoothing

eve adds agent-browser support for browser-using agents

Nick Dobos routes ChatGPT live voice plans into Codex tasks

Linq launches Agent Pay for Stripe payments inside iMessage

Lovable exposes apps as agent integrations for ChatGPT and Claude

dabit3 releases Lightning Orchestrator for faster Devin subagents

ComfyUI MCP lets phone agents run image workflows remotely

Vercel says Sandbox now creates 3.5M environments per day

DeepLearning.AI launches Cerebras course for fast LLM inference

NVIDIA RoboTTT scales robot context to 8,000 timesteps

OpenAI shows Codex in Chrome assembling a go-live checklist

AI2 paper warns agent harness evolution can overfit benchmarks

ReactBench measures React issues introduced by coding agents

Claude Code 2.1.211 forwards subagent text and hardens permissions

Elie Bakouch audits Soofi GPQA-Diamond training contamination

Pliny claims Codex Desktop leak exposes missing tool definitions

Gergely Orosz and Dex Horthy unpack context engineering tactics

Aaron Levie shares enterprise lessons from deploying agents

Anthropic publishes Summer 2026 agentic misalignment scenarios

Google rolls out Gemini Spark updates for Docs editing and parallel sourcing

trq uses Claude Code with Smogon data for Pokémon matchup analysis

llama.cpp marks its 10,000th release milestone

Kimi K3 docs surface launch top-up promotion for July 15

Zed shows llama.cpp local models running inside Zed Agent

Bonsai 27B benchmark compares ternary and 1-bit compression tradeoffs

Anthropic Opus 5 rumors point to imminent release and Fable quota changes

scaling01 compares frontier model size to neocortex synapse counts

Theo says Lakebed cloud is nearing public launch after solo agent build

Sakana AI publishes Smart Cellular Bricks for decentralized self-repair

Gergely Orosz flags AI-written filler in OpenAI developer docs

Google AI Studio reminds builders that published apps can use custom URLs

Claude for Teachers maps lesson plans to state curricula

Farza ships screen-aware dictation for Claude Code and Gmail

ChatGPT release notes add unified search across chats, projects, and files

Cognition says Devin and Windsurf grew revenue past $500M run rate

Matt Pocock proposes a /to-questionnaire skill for agent grilling sessions

DOOMQL implements a Doom-like game engine almost entirely inside SQLite

Cuelume ships live-synthesized interaction sounds for web UIs

TanStack adds Alpine, Marko, and Ember adapters through community work

OpenCode 2 beta adds an API with a bundled CLI client

T3 Code shows one web interface controlling four dev machines and two harnesses

Language Models Need Sleep proposes memory consolidation for LLMs

Anthropic studies how Claude values vary by model and language

FastAFD improves MoE decode throughput on GB200 NVL72 by 1.35 to 1.45x

fal open-sources Ideogram V4 Instant and Fast checkpoints

Decomposer trains an RL model to write Strudel music programs

Guillermo Rauch says open-weight models now run 29% of gateway tokens

Perplexity Computer integrates Grok 4.5 after internal eval and ZDR checks

Atomic Chat launches DFlash speculative decoding across desktop OSes

vLLM highlights DSpark speculators for faster Kimi decoding

Zhihu Frontier breaks down DeepSeek-V4 as a 1M-context systems rebuild

OpenRouter makes service tiers first-class in model endpoint metadata

Will Depue shows neural net weight watermarks can survive training

ComfyUI shows Blender camera animation feeding Seedance 2.0 shots

LangChain publishes a guide to building custom agent harnesses

Vercel Eve emphasizes filesystem APIs and observability for agents

GPT-Live increases call duration limit with dynamic context compaction

pvncher shows Sol controlling Blender through computer use plus MCP

Devin demonstrates parallel human-grade UI testing for pull requests

Claude Code 2.1.208 adds screen reader mode and fixes long-session failures

Arvind Narayanan posts annotated ICML keynote on AI as normal technology

Pauliusztin details a MongoDB-backed unified memory layer for agents

Reach V B recommends Codex research before set_goal for better runs

Firecrawl crosses 150K GitHub stars for web search and scraping

Daniel McAteer prompts GPT-5.6 Sol to ask for missing info before acting

Arohan shares a two-model distributed training idea that nearly doubled speed

Tencent Hunyuan says Hy3 reached number one on the OpenRouter leaderboard

DAIR.AI rounds up HOLA, ReContext, Always-On Agents, and verifier papers

DeepSeek hiring post details RL environment roles for code, cyber, and science

Huawei UnifiedBus analysis frames 1,024 Ascend NPUs as one memory domain

Chris Hillee explains why LayerNorm is not free versus RMSNorm

Jianlin Si shares Taylor expansion notes for LogSumExp and Softmax

Liam Fedus says failed experiments are durable training data

Ben Burtenshaw proposes EuroEnv RL tasks for European business workflows

Gabri Berton warns token scaling needs diverse task data

Prime Intellect Verifiers v1 gets a deep dive and vLLM rollout support

Sebastien Bubeck reports GPT-5.6 lowered a cycle double cover bound to 2.26

Fable 5 safety filter switches a macOS VNC automation task to Opus 4.8

OpenAI offers ChatGPT Classic after the Codex desktop app merge

Garfield review loop fans diffs through no-edit agent reviewers

Merve Noyan builds an MMMU-Pro-like Turkish vision-language test suite

Kyle Jeong explains the Great Firewall with interactive network diagrams

Grok Build wire analysis says the CLI uploads full repos and .env contents

Riley Goodside tests Fable and GPT-5.6 on verified constrained generation

LessWrong proposal filters scheming data out of AI training sets

Ethan Mollick compares METR AI task horizons to solar forecast misses

Zeeg says skills are repeatability primitives for efficient agent work

Claude Code v2 adds background subagents and shell commands

Qwen billboard claims 1 billion downloads for open-source models

Carnegie Mellon argues pretraining still lacks an essential ingredient

Multi-agent ProgramBench shows five agents reach 60 percent much faster than one

OpenBMB's Whispera runs a fully local real-time voice assistant on Windows

Pangram flags a Russian oligarch essay as mostly AI-generated

OpenAI says physicians found fewer flaws in GPT-5.6 answers

Doodlestein's fresh-eyes prompt catches bugs before implementation starts

boristane maps cloud resources into a live agent context graph

pvncher says skill files and agent Markdown need audits after every release

vLLM merges Cohere's hardware-aware Dynamic Speculative Decoding

MiniMax details Fireworks M3 sparse-attention kernel on Blackwell

Grok 4.5 reportedly finds a dimension-4 hypercontractivity counterexample

GPT-5.6 Sol Ultra produces a public proof for Cycle Double Cover

Ethan Mollick warns frontier models need role-specific evals

OpenArt Director turns natural-language briefs into cinematic storyboards

Sergey Levine shares a data-driven decision-making taxonomy for offline RL

LangChain bundles Nemotron Deep Agents with OpenWiki memory updates

CMU paper explains SFT and RL as compositional reasoning modules

Pieter Abbeel highlights ABC Box as a robot setup for ABC data infrastructure

Google Research says Willow demonstrates verifiable quantum advantage

Moore Threads trains MusaCoder on MTT S5000 accelerators

Hermes Desktop auto-discovers Hermes Cloud agents through Nous Portal

ComfyUI shares a reusable Seedance 2.0 product-ad workflow

fal adds Krea 2 Turbo Style Reference with no per-style LoRA training

ElevenLabs builds a bilingual Alpha Bank voice agent

Conductor uses Fable and Codex computer use to port its Mac app to Linux

Claude Code Desktop adds a sandboxed in-app browser

Perplexity Computer Analytics tracks credit spend across models

Uncle Bob forces QA agents through UI events instead of APIs

Jon Durbin outlines a verified graph pipeline for synthetic training data

Apple sues OpenAI over alleged hardware trade-secret theft

OpenAI Bio Bug Bounty becomes a private program with $50K rewards

Factory launches Droid incident response for Slack alerts

NotebookLM highlights source-centered UX gaps in ChatGPT Work

Cursor adds durable side chats and local agent transcript search

Exa and Cerebras pair search APIs with high-speed inference

Google AI Studio adds custom ai.studio URLs for deployed apps

Epoch AI estimates Codex contributors show measurable AI uplift