Where people deep in AI come to stay current.
Category
Tags
ChatGPT is your AI assistant for writing, coding, learning, and more.
Open-Source AI Agent That Grows With You
Your personal AI assistant.
A coding agent for the terminal.
Chat answers. Agents do tasks. Computer works.
ChatGPT Work is an agent that can take action across your apps and files, stay with a project for hours if needed, and turn a goal into finished work.
Bring Grok to your computer.
The API for web data
Modal bills sandbox resources each second, subject to a requested minimum, while bursts use spare host capacity. Modal says typical sandboxes spend over half their lifetime below 6% of peak CPU use.
OpenAI's healthcare computer-use agents are available through its API. The update also adds clinical search improvements and more than 20 plugins to ChatGPT for Healthcare.
OpenCode Browse 0.1.0 removes Playwright and claims 25 times lower memory use. Its browser actions compose with MCP tools in codemode and control the user's existing browser.
Simon Willison prototypes external sandboxes for local coding agents
BadEvolve uses poisoned tasks to implant backdoors in agent skills
EvoTrias separates planning, coding, and scoring in program search
LangChain rebuilds LangSmith around a governed agent lifecycle
Qdrant 1.20.1 speeds up search and adds agent integrations
Omar Sar compiles 2026 papers on agent evaluation methods
Kuncheng Gu proposes a restricted macOS account type for agents
Sebastian Raschka calls for a modular API for LLM architecture experiments
Guillermo Rauch argues agent traces should preserve intent alongside code
Plannotator Inbox adds prototypes, rationale, and PR-linked decisions
Robert Martin runs Grok Bot jobs across two local computers
Mastra Connect adds Microsoft 365 apps as agent providers
OpenAI doubles its ChatGPT academic researcher program to 200,000
Matt Pocock charts a year of AI coding workflows and their tradeoffs
Orosz reports compromises at five infrastructure providers
Claude Code adds iPhone Duo simulator support in public beta
Perplexity launches Monitor API at $1 per 1,000 scheduled runs
Cua reports 97% non-interference on Omarchy in a seven-task-per-OS test
Unsloth adds beta computer use for local models
Ollama and Together explain how coding agents preserve prefix caches
ParseBench-2 adds adjustable error penalties in Metric Lab
PyBot creators say 1,000 DeepSeek agents built a chatbot in one week
Claude Code team publishes its verification approach
OpenAI offers free frontier-model access to up to 200,000 scientists
AIE Code talk proposes restrictive APIs to reduce coding-agent mistakes
Perplexity launches Frontier Research with a commitment of at least 10,000 GPUs
Swamp turns repeated agent skills into reusable workflows
Conductor 0.95 adds agent orchestration for local workspaces
Lovable gives app drafts separate backends
OpenClaw 2026.11.4 fixes Signal and WhatsApp message-delivery failures
Chandra 2.7 raises OmniParseBench table score from 91.6 to 94.2
Report: Hugging Face reportedly releases 44.7 TB robotics dataset with 10,000 hours of video
Noeri launches with $110 million to build self-improving AI systems
Google AI Studio showcases audio models for speech and sound generation
Cursor sends dictated voice prompts straight to its coding agent
Google expands Antigravity science access to Gemini 4 Carbon
Baseten previews a hosted voice-agent stack with sub-second turns
Google Antigravity’s /swarm gives agents roles on a shared message board
HUB releases HubEgo-100K workplace video data under CC BY 4.0
Google lets Managed Agents run tools inside customer compute
Omar Khattab argues AI researchers should invest in projects, not just papers
Artificial Analysis benchmarks time-to-first-audio across speech models
LangChain outlines goal engineering for agents that optimize objectives
Satya Nadella says teams should separate agents from their harnesses
A software-factory debate shifts toward shared agent workspaces
Agent automations need an SLA and an accountable engineering owner
Vespa adds agent-ready schemas to improve search query quality
Apple updates macOS full-disk-access prompts for AI assistants
Agents build an HTML5 game by collaborating inside an email thread
S2 builds authoritative DNS against its live basin database
AI helps accelerate llm.c experiments but struggles with algorithmic leaps
Matt Pocock builds a dated LLM wiki from personal communications
A coding chief-of-staff agent coordinates subagents like a merge queue
Glass Health adds a clinical agent builder and new API endpoints
E2B adds Hatchet and Anchor Browser agent cookbook integrations
Omnigent starts agent sessions from Slack, Jira, GitHub, and PagerDuty
Liquid AI releases d1.2 decision models with improved index scores
Vatsal Trivedy argues persistent agents need goals, not task prompts
Claude helps verify theorems hidden in Gödel’s notebooks
Armin Ronacher makes the case for JavaScript as an agent tool
Together raises agent task completion from 67% to 96% with evals
LiveKit leads a five-run voice-agent reliability comparison
A CMU study finds planning agents should show options early
Vincent Koc previewed an iPhone automation package that reads accessibility XML alongside vision. He reports 2 ms latency in an 18-hour multilingual keyboard test, but the package requires a host computer.
Zeeg is prototyping a Pi-based email assistant that filters an existing inbox through APIs and learns from curation. Hosting remains tentative while encryption requirements are unresolved.
CodeClimb’s builder describes Inngest checkpoints that preserve coding experiments through pauses of up to three days. Separate coding and evaluation sandboxes use scores and reviewer feedback to accept or revert changes.
QuixiGate ships as a cross-platform Rust binary for routing requests to cloud and local models. It supports OAuth subscriptions, telemetry integrations and configurable fallback when usage limits are reached.
Gergely Orosz’s analysis identifies missing database-node monitoring and fragile quorum placement in Revolut’s five-hour card outage. He also questions recovery tools that depended on the failed GCP zone.
ChatGPT Wallet lets Dots complete approved purchases without exposing card details to agents. The rollout starts with US Pro subscribers through partnerships with American Express, Mastercard and Visa.
Claude Code’s Projects coordinator now answers questions and reports parallel-thread status in project chat. Updates also add reminders, read-only sharing, /btw and /fork support, and easier artifact navigation.
Vik Paruchuri argues that ParseBench-2’s page-number requirements reward parser defaults rather than OCR quality. The creators defend the open methodology while acknowledging annotation and metric issues and updating the leaderboard.
LMSYS reports that Vera CPU sandboxes reduced average agentic RL episode time by 11% in Miles tests under load. Command latency rose 5–22% as concurrency increased from one to 176 sandboxes per socket.
Antigravity's new Orchestrator uses Carbon to plan and Hydrogen to edit and test. Google reports near-Carbon Terminal-Bench success with 24% less quota use and 19% faster completion; rollout starts with AI Ultra subscribers.
Get the best stories deliveredto your inbox