The AI model wars have a new front.

Instead of fighting to build the best model, Perplexity just asked a different question: what if you didn't have to choose?

This week's move is less about capability and more about control.

Perplexity just launched Perplexity Computer a multi-model orchestration system that routes tasks across 19 separate AI models from competing labs, all under one workflow. Users describe an outcome, and the system spins up sub-agents that browse, code, connect to apps, and run autonomously for days, weeks, even months inside isolated sandboxes. Pricing is consumption-based, with the Max tier getting a 10K-credit monthly bank.

This is the first real attempt from a major AI player to make model flexibility a core product feature - not just a backend toggle. Rather than winning by building a better model, Perplexity's bet is: use the best model from everyone, for every job.

CEO Aravind Srinivas made the competitive angle explicit, writing that "the biggest weakness of Claude is that it only coworks with Claude."

On X, user @hamptonism used Perplexity Computer to build a Bloomberg Terminal lookalike in a single session. The result, dubbed "Perplexity Terminal," pulls real-time data from Perplexity Finance and renders a full dashboard with tabs for Overview, Price Chart, Analysis, Earnings, News and more all pointed at NVIDIA stock. Bloomberg Terminal costs roughly $30,000 a year. And he did it in one prompt.

To be fair, the community pushed back fast. Bloomberg's real edge isn't the interface. It's decades of proprietary data across fixed income, derivatives, and institutional workflows that Yahoo Finance simply can't replicate. Martin Shkreli's response was blunt: "Not another one of these lol." Fair point. But the more interesting signal isn't whether this beats Bloomberg. It's that someone built a credible-looking financial terminal in one sitting, for free. That's the demo that matters.

This is a smart play from a company that can't win the model race. Perplexity is competing at the orchestration layer instead, and if users start expecting AI to automatically route tasks to the right model, the company that owns that experience won't need to win on training.

The real question: does multi-model actually perform better, or is this a narrative play for a market tired of model benchmarks?

You don't have to read everything — just the right thing. 1440's daily newsletter distills the day's biggest stories from 100+ sources into one quick, 5-minute read. It's the fastest way to stay sharp, sound informed, and actually understand what's happening in the world. Join 4.5 million readers who start their day the smart way.

Join for free today!

A new survey found nearly 1 in 8 American teenagers are going to ChatGPT, Claude, and Grok for mental health advice and emotional support. The problem: none of these tools were designed for it, and mental health professionals are raising flags about the risks of leaving the most vulnerable users the least protected.

→ AI is filling a gap society hasn't solved, but no one's measuring the cost yet.

After spinning out in 2021, Intrinsic, Alphabet's industrial robotics software unit, is being absorbed back into Google. The move signals that as AI robotics heats up, Big Tech is consolidating firepower rather than spreading it across subsidiaries.

→ The robotics race is forcing internal mergers, not just external acquisitions.

Harper, an AI-native insurance brokerage from YC's Winter 2025 cohort, just closed $47M across a combined seed and Series A. It's one of the clearest signals yet that vertical AI in financial services is attracting serious capital, not at the model layer, but at the application layer.

→ The next wave of AI winners might not be building models. They're building industries.

If you're on Perplexity Max, try this: create a Space with the instruction below, then set it as a scheduled daily task:

"Every morning, find the 3 most important articles about [your industry] from the last 24 hours. Summarize each in 2 sentences, rate relevance 1 to 5, and explain why I should care. Output as a clean table."

Use it for: competitor monitoring, industry tracking, newsletter prep, client briefing. The difference from ChatGPT: this runs in the background. No tab needed.

It's already done when you wake up.

Perplexity isn't trying to build the best model. They're building the layer that sits above all of them. If that bet lands, the AI competition in 2 to 3 years won't be about who trained the smartest model.

It'll be about who controls how people actually use them.

👉 Forward it to someone trying to keep up with AI.

Read online: leftcurve.beehiiv.com