【AI Industry Alert】After OpenClaw, Perplexity has dropped another bombshell. Perplexity Computer is fully armed with 19 top AI models, all working together without any human intervention. At the same time, Anthropic quietly made a major move, with Claude executing the most critical acquisition in its history.
Say goodbye to OpenClaw. The era of forced equality is over.
Just yesterday, Perplexity quietly released Perplexity Computer, a system fully armed with powerful AI.
It is not just a chatbot. It is a complete AI agent system that can handle research, writing, coding, calculations, and any project.
More importantly, Perplexity Computer is a multi-model system. Through intelligent scheduling, multiple AI models work together and supervise each other.
The core logic behind it is: for different tasks, use the most suitable model, with a total of 19 models.
The system automatically selects the best model for each task. For example, Claude handles research, Gemini handles speed, Grok handles background information, and small models run in the background 24 hours a day.
While Perplexity is betting on the agent swarm model, Anthropic also made a big move.
On the same day, Anthropic officially announced the acquisition of Vercept, a company focused on the future of Claude.
ai furry porn
Perplexity Computer Attacks with 19 AI Models
Perplexity Computer is the answer to what people in 2026 need most from their computers.
sextbots
Because it is an AI system, it can remember the user’s past work and automatically complete tasks safely and reliably.
The system runs in a sandbox environment, has file system permissions, and can directly access files and client data.
porn ai chat
Whether it is writing a novel, creating a game, or any project, you can create a new project and Perplexity Computer will fully take over the task.
Moreover, it does not just complete one task. You can assign multiple different tasks at the same time and push dozens of projects forward together.
Perplexity scientists say that because it is built on a large model, OpenClaw is also very “intelligent” and “autonomous.”
More importantly, Perplexity Computer’s electronic brain is given a completely new form: the full agent swarm mode.
What you need to understand is “hero stock value analysis.” Computer can go to AMD, AVGO, INTC, QCOM, TSMC and other company websites, collect data, and display it in a web application or slide.
Perplexity Computer can also create a Dario Amodei-style podcast, with model selection, chip analysis, and more.
AI agents can also create viral short videos suitable for TikTok, automatically generating scripts, images, and videos.
For example, create a Walter White-style video, and then analyze the housing prices in Chengdu.
Or track Airbnb’s super host investment return rate, and Computer can use multiple models to complete the analysis.
What is worth mentioning is that Perplexity Computer strongly implements a concept that a former GitHub Copilot architect secretly revealed.
This person is Alex Graveley, the former chief architect of GitHub Copilot.
After leaving Perplexity, he founded Comet, an AI-native company.
Currently, Perplexity Computer’s web version Max has a waiting list, with 10,000 people per day.
Pricing model: users can choose the “smartest” or “fastest” mode for different tasks, or choose a specific model, with no token limit.
As a gift, users who sign up now will receive 20,000 tokens for free, valid for 30 days.
The Founder Reveals the Core: AI Is the New Operating System
After the release of Perplexity Computer, founder Aravind Srinivas published a long post revealing the four major changes in the AI era.
He bluntly pointed out that the biggest flaw of a single model is the lack of collaboration. The model itself has become the biggest bottleneck.
Among them, the first change is to regard internal tools as “ASI” practical experience.
He gave an example: a Slack bot that can show local weather.
But a real agent can create a complete project, write code, create documents, and connect to multiple APIs. It can even write a novel in the style of Liu Cixin.
AI is no longer a simple chat box, but a file system with Shell permissions, a real “computer.”
More importantly, Srinivas revealed the historical shift in thinking: in 2011, Google launched Chromebook, which simplified the computer to a browser. Now, AI is turning the browser back into a computer.
He believes that future computers will no longer have GUI applications. 19 AI models will form a “task force.”
Claude handles research, Opus handles coding, and different models are used for different tasks.
More importantly, the “intelligent operating system” should be built with multiple models, using the strongest model for logic.
Srinivas said that in 2025, on average, one new model is released every 17 days. The current situation is like the bottom layer of the river, where the water flow is turning into the electrical signals of structured models.
He calls it “agent swarm mode,” which indicates that the traditional OS foundation is being rebuilt by an AI operating system. The time has come.
One App Kills the 30K Dollar Terminal
Perplexity is not just talking. It has fully implemented what it promised.
Users can choose their preferred model: Claude, Gemini, or GPT.
This is not just a toy for tech enthusiasts at the cutting edge of the traditional terminal.
Through model collaboration, Perplexity Computer can completely replace Bloomberg’s “real-time data terminal.”
In the demo, it can automatically collect stock data, generate performance charts, company information summaries, and market trends.
You need to know that Bloomberg terminals cost as much as $30,000 per year, relying on real-time data and expert analysis.
Perplexity bluntly said that Bloomberg’s monopoly position is about to be challenged.
What is even more shocking is that Perplexity Computer also created a web application that displays the real-time positions of all satellites.
It grabs data from CelesTrak for 10 satellite constellations, about 1,500 satellites, and their orbital elements (TLE).
Then it uses satellite.js and the SGP4 algorithm to calculate, updating once per second, converting each satellite’s orbital parameters into real-time position.
In the entire research process, Computer used multiple different models.
Claude generated the GitHub PR, Gemini explored the data, and Opus wrote the code.
Zero cost, fully detailed hero stock value analysis.
Anthropic Acquires Vercept to Strengthen Claude’s Computer Use
While OpenClaw is being replaced by the new generation, Anthropic is also quietly strengthening Claude’s computer control capabilities.
Yesterday, Anthropic officially announced the acquisition of AI agent company Vercept.
This acquisition has an extremely clear goal: to make Claude truly usable. “Computer Use” is no longer just a demo, but a real thought.
Current AI agents are like a bottle that is half open. They cannot fully open a known and solved problem.
Vercept is the key to opening it.
The company was founded in 2024 by former Meta and Ai2 researchers Ross Girshick, Luca Weihs, and Kiana Ehsani.
Before the acquisition, Vercept had been building AI systems that can understand the “daily life” of computers, observing and acting.
From a technical perspective, Anthropic currently has the most complete computer control capabilities in the industry.
Last week, Claude Sonnet 4.6’s “computer use” capability shocked the entire industry.
On the OSWorld benchmark, it jumped from 15% at the bottom in 2024 to 72.5%, far exceeding the average human level.
In the future, Vercept will be fully integrated into Claude.
After acquiring Bun, Anthropic once again expanded its territory through acquisition, sending a clear signal: the key to future AI agents is whose “hands” can be used.