claude now runs on the google cloud login and invoice you already have, new stuff landing day one. the boring plumbing launches are my favorite kind, nice work team
cognition open sourced the fix for this week's php and node bugs for whatever coding agent you use: it finds the affected projects, patches them, tests them, hands you one change to approve each. "please fix my php vulnerability" works now, nice work team
We’ve open sourced an agent plugin that helps secure systems against recent cybersecurity threats.
You can install them with any agent or use them in Devin today:
devin.ai/blog/vulnerability-…
canva trained their own image model from scratch, out today in dream lab and their api. it plans the layout before it draws, so text in the generated image actually reads, the part every other model faceplants on. i'd use it for deck slides alone
Introducing Canva Craft 🚀
Extraordinary image generation is now available to everyone, powered by our first frontier model built specifically for design.
With Craft, Canva is now the world's second most advanced AI lab for image generation on @DesignArena's global leaderboard.
See what it can do 👇
liquid ai shipped a serving engine that answers a whole request with a small model in under 50ms end to end, fast enough to sit in the live search and ads path where only boosted trees used to fit. shopify's already running it sub-20ms in prod
Today, we announce 𝙻𝚒𝚚𝚞𝚒𝚍 𝚁𝙸𝙿𝙻 (real-time inference pipeline), our ultra low-latency inference engine built to put LFMs directly in the request path of real-time workloads like search, ads, fraud, recommendations, and payments.
> complete generated responses end-to-end in sub 50ms
> maintains low latency under continuous peak load
> LFMs get up to 20x faster on RIPL
> RIPL supports Multimodal LFMs
> RIPL supports d1, our decision models too!
RIPL optimizes the full path from request arrival to a complete generated response. As traffic rises, its performance advantage is greatest on the slowest requests. Let’s see some production use cases:
google's handing every AI agent its own tiny computer, and it freezes the second the agent stops to think so you're not paying for it staring at a wall. i'd steal that
Announced at AGNTCon + MCPCon: the new Agent Compute Engine plus expanded Gemini for business agent governance deliver the control you need to run, govern, and scale your agents.
Learn more about Agent Compute Engine and expanded agent governance → goo.gle/4yFN3yT
kevin weil (ex openai science) raised $215M for Warp Sciences: a robot lab in the US that makes and tests hundreds of versions of a molecule, the step pharma keeps shipping to china. chemists in there wear sensor gloves all day so the models learn from their hands
💥 Introducing Warp Sciences, a new company applying AI and robotics to help the world develop new medicines faster.
We’ve raised a $215M Series A led by @lightspeedvp to bring frontier AI out of the browser and into the lab, and build scientific capacity here in America.
I co-founded Warp with my long-time friend @dpatil, FAIR co-founder @manohar_paluri, and drug hunter Sol Reisberg. The early team is full of incredible people like @bkspears9, who helped achieve Q>1 fusion at LLNL.
firecrawl just put 30M+ pages of US law, statutes, court opinions, city zoning and permit codes, behind one search you can call from chatgpt or claude, free. the municipal permit stuff was always buried in garbage pdfs, nice work team
Introducing the free Government & Legal Index for ChatGPT & Claude.
Give your AI agents 30M+ US government pages, from federal statutes and court opinions to city codes and permits.
Research tax rules, permits & case law with the Firecrawl plugin.
firecrawl.dev/alexandria/gov…
congrats team, prompt in, real svg out, and it opens in figma with actual curves you can drag. that's the part of logo work nobody signs up for
Today, we’re introducing Ideogram's first vector model.
Generate clean, editable SVGs with smooth gradients and sharp typography from a text prompt, or turn any image into a detailed, scalable vector.
Live in Ideogram and the API: ideogram.ai/models/vector
1password now lets your agent ask for a login, you approve on your device, and 1password types the password in itself so the agent never sees it. free for personal accounts. want to know if the agent can just read the page after that
Today we announce 1Password for Agents platform, building with companies @Muse, SpaceX Super Intelligence Grok @Bot, @AnthropicAI’s @Claudeai, @OpenAI, @TownAI, @Browserbase, @Browser_use, @heydana_ai, Instinct, @usekernel, @Perplexity_ai, @tryramp, @Tomo, and @wajo_ai, with more integrations on the way. 🎉
We've heard from so many of our customers who want to use these new personal AI agents to help them get more done, but want more control over what the agent can access with the same trust and ease of use they get when using 1Password directly.
1Password for Agents gives your agents access without giving up your passwords. When an agent reaches a sign-in form, it requests a credential from your vault. If you approve the request, 1Password directly auto-fills the credential in the destination site so the agent can’t see or take possession of the credentials.
The agent gets the task done, but it doesn’t get your passwords..
You stay in control. 1Password for Agents free and optional for personal and family accounts. Unless you opt-in to an agent connector, agents cannot touch your vault.
➡️Learn more: bit.ly/487da6J
deepmind and WRI sorted every 10m patch of earth into real forest vs tree farm. every "we planted a million trees" brand post is about to get audited. nice work team
llamaindex shipped OCRia-1, a little 4B model that reads ugly pdfs and turns tables, charts and formulas into clean data, beating parsers twice its size. nice work team, small model winning is my kind of launch
Today we’re releasing OCRia-1, a frontier VLM for document parsing that sets a new SOTA in accuracy, cost, and latency.
It not only sets a new high score on ParseBench (87.4), but also leads in three 3rd-party benchmarks: olmOCR-bench* (83.5 overall, 92.9 on tables!), PulseBench-Tab (84.1), and fr-bench-pdf2md (72.1).
OCRia-1 is really, really good at parsing complex tables: nested headers, merged cells, irregular grids, scanned tables, and more. It is also fantastic at parsing complex unlabeled charts (line, bar, pie, technical diagrams), and preserving complex formatting. Most real-world enterprise documents contain at least one of the these, whether it’s a big tax withholding table or a public economic report.
It is a 4B model and outperforms all other models within a similar parameter class: LightOnOCR-3, Chandra-OCR 2, Mistral OCR 4.1, and more. It is also extremely fast.
All credit goes to our research team, who spent countless hours on curating a diverse dataset, tuning our training pipeline, and squeezing every last drop out of model optimization so that it can process billions of pages at scale.
OCRia-1 as a base is powering all of our modes within LlamaParse (cost-effective, agentic, agentic plus), so you can access its capabilities today!
If you’re processing millions/billions of documents at scale, we’re offering enterprises that book a demo up to $3k in credits for a guided evaluation.**
*olmOCR-bench scores are adjusted based on unifying rendering at 1,288px for all models
**enterprise POC credits apply to eligible organizations that book a demo
google's AI studio shipped a cli that hands the gemini api to whatever agent you already run: make the picture, the voiceover, the song, the video, or read a pdf. no throwaway script per call. one binary, nice work
axios says banks can't tell a bot from you, so the plan is agents get to look at your money and not move it. morgan stanley just skipped to blocking them
SCOOP: Major U.S. banks are detecting unauthorized third-party AI agents inside customer accounts, including some attempting to take actions that appear to be independent of user instructions, sources told Axios.
The findings have alarmed banking executives, who are racing to restrict access to their systems as AI moves at a velocity faster than financial institutions can respond, multiple industry and bank executives told Axios.
Several banks are now preparing for worst-case and even “day after” scenarios, including how to rapidly secure or restore customer accounts if things go wrong, the people said.
Several bank executives told Axios they're struggling to distinguish between human users and AI agents. One executive said banks also struggle to determine whether agents are authorized by customers or potentially operating without their permission, more akin to rogue agent behavior.
Read my latest for Axios here: axios.com/2026/10/22/ai-agen…
scenario open sourced a Blender plugin that makes 3D, textures, video and audio from a prompt right in the sidebar, and the demo genuinely turns "harbor town" into boats and a lighthouse. code is free, the generations burn paid credits, that's the whole catch
we just open sourced the last @Blender AI plugin you’ll ever need
github.com/scenario-labs/ble…
> 500 models, one tab. images, video, 3D, audio, textures, worlds
> built for humans and agents
> live now
playasia, the japanese game shop people have been importing from for 24 years, just took the whole store offline to patch after this week's PHP attacks. nobody wants to go dark in october, but that beats leaking everyone's orders
We have temporarily taken the site offline.
This is Jan, the real CEO of Playasia.
After weeks of increasingly sophisticated and relentless attacks against us and several of our trusted partners, we made the decision to go dark while we complete a full security audit and harden our systems.
This is not a step we take lightly. Playasia has been our home, and your home, for nearly 24 years, and the thought of stepping away even for a short while breaks my heart.
These attacks are not random. From what we have seen in the attackers' own source code comments, we are listed as one numbered goal among tens of thousands of other targets. That tells us this is a coordinated, systemic campaign, not something aimed only at Playasia.
Watching our team work around the clock, exhausted but still fighting to protect what we have built together, has brought me to tears more than once. Many of them have families who depend on them. I have my own. The passion that has driven this community of gamers serving fellow gamers for over two decades is something we will not let anyone take from us.
We have reported everything to the relevant authorities and cybercrime units worldwide and continue to share every detail that can help them pursue those responsible. We are doing everything in our power to protect our customers, our partners, and our people.
Your support over the years has meant more than I know how to say. Please stay tuned for updates on our social channels and here.
We will be back.
Jan
tencent built 16k practice environments out of real office files and repos, then grades the agent on whether the writeup actually says the right thing, not just whether a file appeared. that's the part every agent demo skips, nice work team
📢 Introducing WorkForge: a scalable framework that constructs verifiable training environments for long-horizon work agents from real-world resources.
Starting from expert workflows, it builds authentic workspaces, extracts checkable facts, and derives tasks plus dual (programmatic + semantic) verifiers grounded in observable evidence.
🚀 We scale to 16.7K environments across 40 professional domains covering 60 file types, enabling consistent gains on long-horizon work benchmarks as data volume and interaction horizon increase.
📚Paper: arxiv.org/pdf/2610.04906 @zhang_leo54389 @ZoeyyyZhououou @Be1ong1
#Tencent Hy#AIAgents #WorkAgents #AgentTraining
hugging face released Carbon-X: paste raw DNA into a chat and a 14b model answers in english, benign or pathogenic, with reasoning. within a point of the frontier model at a tiny fraction of the cost, and it fits on one gpu so your sequence stays yours. okay this is cool
Today we're releasing Carbon-X, a 14B model that brings DNA and natural language together for the first time 🧬
1. Beats every open language model we tested across DNA benchmarks. Within a point of GPT-6.1 Sol at >1,000× lower inference cost.
2. Mid-trained on 300B tokens of interleaved DNA and text, followed by SFT (without distillation!) and RL.
3. A new hybrid tokenizer encodes DNA using ~3× fewer tokens than BPE.
4. Gives other models a DNA specialist subagent: Kimi K3 gains 11.1 points when it consults Carbon-X.
5. Runs locally on a laptop with llama.cpp. Apache 2.0. Your DNA stays on your hardware.
Model: huggingface.co/HuggingFaceBi…
Blog: huggingface.co/spaces/Huggin…
ant group's ling team shipped another flash model, 44 on the index people rank these things with, up from 41, and the agent jumps are the actual story. the open weights aren't out yet and that's the part i'm waiting on
