Discover the Best AI Tools

We test the top AI tools for writing, video, images & music — so you don't have to.

Explore AI Tools

Cursor Just Revealed How a Leaked Vendor Key Led to a Tampered Claude Code Extension

-
Image
  Cursor has finally explained how its October security incident really happened — and the trail leads outside Cursor itself. In an update published on November 13, 2026, the AI coding company said the breach began at Fireworks AI, one of its inference vendors, after an access key it had issued to Fireworks was exposed there. What followed was a chain reaction: stolen coding agent traces, harvested credentials, and a tampered extension slipped into the cache that serves Cursor's marketplace. Here is what happened, what was affected, and what you should do now. How the breach unfolded According to Cursor, the attacker used the exposed Fireworks key to copy coding agent traces from a read-only storage bucket Cursor used to share data with Fireworks. Most of the traces dated from 2025 and came from sessions outside Privacy Mode. A small number of those traces contained unredacted credentials. Cursor says fewer than 0.05% of users had traces exposed, and fewer than 0.01% of customers h...
Post a Comment

OpenAI Just Launched GPT-6.1 Sol — Near-Flagship Smarts at One-Fifth the Price

-
GPT-6.1 Sol launch cover — near-Astra performance at one-fifth the price

OpenAI logo

What just happened

On the biggest stage of its year — the DevDay 2026 keynote in San Francisco — OpenAI launched a new AI model called GPT-6.1 Sol. It's not the flashiest thing announced at the event (that was Dots, the always-on agents), but it might be the most useful one for anyone building with AI.

Near-flagship brains, fraction of the bill

GPT-6.1 Sol sits between the workhorse GPT-6 Luna and the flagship GPT-6 Astra. OpenAI says the new model comes close to Astra on demanding tasks like coding, debugging, understanding documents, and multi-step jobs — while costing about one-fifth per task. On the DeepSWE 1.1 coding benchmark, GPT-6.1 Sol matched Astra while running roughly five times cheaper. For the average person, think of it as first-class answers at an economy price.

Fewer made-up answers

OpenAI also reports the new model hallucinates less. When using its "low reasoning effort" setting, factual mistakes fell from 11.4% to 7.7%. Its error rate is now within 1.9 percentage points of Astra across reasoning settings. It also claims to fail less often when told to stay within limits, flag broken tools, and avoid doing things it wasn't authorized to do. Coming right after OpenAI shelved its even bigger GPT-6.1 Astra over safety concerns, this feels like a deliberate message: speed and smarts, but safer.

Where you can get it

Starting today, GPT-6.1 Sol is available to Plus, Pro, Business, Enterprise and Edu users in ChatGPT Work and Codex. Developers can use it through the API as gpt-6.1-sol. It's not in the regular chat window yet. OpenAI also teased an "Ultrafast" version coming soon that should generate tokens up to eight times faster inside Codex. API pricing is $2 per million input tokens, $10 per million output tokens, and just $0.10 per million cached input tokens — a 95% discount.

The take

This is the price war in action. Anthropic just made Claude Sonnet 5.5 faster and cheaper; OpenAI is answering with a model that's nearly as good as its best at a fraction of the price. For beginners, the practical upshot is simple: better AI tools are about to cost less — whether you use them directly or they power the apps you use.



Comments

Post a Comment

Popular posts from this blog

Anthropic Just Dropped Two New Claude Models — Fable 5.2 and Mythos 5.2

-
Image
  Anthropic has unveiled two new additions to its Claude lineup — Claude Fable 5.2 and Claude Mythos 5.2 — launching on November 9, 2026. The pair arrives as the company's latest push to stay ahead in the fast-moving race for enterprise AI, and both models are built on the Claude 5.2 family, promising sharper reasoning, faster outputs and lower costs than their 5.1 predecessors. What Fable 5.2 brings to coding and agents Fable 5.2 is the general-purpose workhorse of the pair. Anthropic positions it as its strongest model yet for coding, reasoning and autonomous AI agents, and it is broadly available to developers and businesses through the company's usual channels. Whether it is writing and debugging code, planning multi-step tasks or powering agentic workflows, Fable 5.2 is designed to handle long, complex jobs with fewer mistakes and clearer explanations along the way. Mythos 5.2: built for cybersecurity and life sciences Mythos 5.2 takes a different path. It is tuned for cyb...
Read more

AI Tools Every Accountant Should Use in 2026

-
Image
Accountants don't lose time to math — they lose it to the pile around the math: receipts nobody sent, transactions nobody coded, client emails that need three follow-ups, and tax questions that eat an afternoon. AI won't replace the judgment call, but in 2026 it takes the pile off your desk. Here's the stack that earns its keep. Start with the ledger: QuickBooks or Xero Both big ledger platforms went all-in on AI this year. QuickBooks Online ships with Intuit Assist bundled into every plan — it auto-categorizes transactions, drafts invoice reminders, and on the Plus plan and up adds AI-powered bank reconciliation. After Intuit's August 2026 price increase, plans run $38 to $340 a month. Xero answers with JAX — auto bank reconciliation in open beta, Smart Document Capture that reads invoices into the ledger, and a chat that answers questions from your own books. Xero also raised prices October 1 (Early $27, Growing $59, Established $97), but every plan includes unlimited...
Read more

Top 5 Best AI Video Generation Tools in 2026

-
Image
A year ago, making a video meant cameras, editing software, and hours of work. Today? You type a sentence and AI generates the footage. I've been testing video generators for my own projects, and honestly, the results blew my mind. Here are the 5 best AI video generation tools in 2026 — the ones I'd actually recommend to a friend. 1. Runway Best for: Professional-quality video with editing control Runway is the tool the pros actually use. I generated a product-style clip from a single image and the motion looked genuinely cinematic. What sets it apart is control — you can edit specific parts of a clip, extend videos, and fine-tune styles. If you're serious about AI video, start here. Price: Free credits to start; paid plans from $15/month. Try Runway → 2. Pika Best for: Fast, fun clips — great free tier Pika is where I tell beginners to start. It's fast, it's playful, and the free tier is genuinely generous. I made a stylized 5-second clip f...
Read more