Mercor

Agents are only as good as the environments behind them. At Mercor, we've built deep expertise in the realistic, economically-grounded environments that help agents bridge the gap from the lab to real-world usefulness. We want to put that expertise to work for the broader ecosystem—so we're glad to be joining the OpenEnv committee, alongside Meta PyTorch, NVIDIA, Prime Intellect, Hugging Face, and others, to help guide the open foundation for agentic environments. Read more at the link in the comments.

Claude Fable 5 lands 2nd on APEX-Agents, and 1st in Law. APEX-Agents is our benchmark for long-horizon professional work: multi-step, tool-heavy tasks across investment banking, corporate law, and management consulting. Anthropic Claude Fable (Max) scored 45.0% Pass@1, behind Gemini 3.5 Flash and ahead of Claude Opus 4.8. What stands out is efficiency. Fable reached 2nd overall while using 70% fewer tokens than leaderboard leader Gemini 3.5 Flash and 37% fewer than GPT 5.5, in far fewer steps. Across 4 runs it solved 246 of 480 tasks, including 8 that no other model has solved. All 8 are in Law, spanning Family Law, Consumer Protection, Corporate Tax, and Securities. View the full leaderboards: https://lnkd.in/gx2Q_JHi

Anthropic Claude Fable 5 takes #1 on APEX-SWE, our benchmark for real-world software engineering work. It scores 65.5% Pass@1 overall. Fable 5 also tops both APEX-SWE categories: Integration at 61.3% and Observability at 69.7%. Observability is the result that stands out. Fable 5 is 26pp ahead of Opus 4.8, the first model to clear 50% on the category, and the only model that scores higher on Observability than on Integration. Observability has been the bottleneck for every model we have measured, and Fable 5 is the first to break it. What changed is how it works. Where earlier models read code one file at a time, Fable 5 investigates in parallel, running multiple lines of inquiry at once: searching code, reading files, running tests, and querying logs together. It reads surgically rather than by brute force, cutting full-file reads by 91% in favor of targeted commands that pull only the lines it needs. And it spends less effort searching and more validating, testing as it goes, and rewriting less. The pattern is higher leverage, not more effort. Fable 5 uses 10% fewer tool calls and 36% less reasoning text than Opus 4.8, and still comes out well ahead. Congrats to the team at Anthropic on such a strong model release!

AI progress requires (1) compute, (2) algorithms, and (3) data. - The leading compute company is worth $5 trillion. - The leading model company is worth $1 trillion. - Mercor is the leading data company and is currently valued orders of magnitude lower. There's an opportunity in how the market is mispricing the value of data. Data is the oil of the AI revolution. It is the primary way that models and enterprises build competitive advantages.

The best way to understand what it's like to work with Mercor is to hear from the people doing the work. Across projects and domains, our experts keep pointing to the same things. They love the flexible schedules, the intellectually challenging work, the competitive pay, and a platform that respects their expertise and their time. Here's what some of them had to say.

Mercor is organizing human intelligence to power the AI economy. We have over 50 full-time roles open across teams and offices. Join us. Apply today at the link in the comments.

I told Harry Stebbings on 20VC that Mercor now spends more on tokens for our internal agents than we do on headcount. In a few years, every enterprise will too. We already see that the companies pulling ahead are running like AI labs: an eval for every workflow, agents trained like you'd train an employee, models swapped in and out on price and performance. This is what organizing human intelligence looks like. Watch the episode: youtu.be/CkAmq2gYtA4

Companies will spend more on tokens than they do salaries very soon. Application layer companies have no defensibility, the model is the product. Hiring researchers will cost you tens of millions of dollars today. Everything you think you know about defensibility, token spend, labour displacement, will be changed following this discussion. I condensed the core ideas which changed my thinking in my conversation with Brendan Foody @ Mercor below. 1. Why Frontier AI Labs Could Become $10TN Companies Critics once questioned whether frontier labs could keep pricing power in a competitive market. Their revenue velocity now suggests the opposite. The opportunity around leading frontier models is so large that it could absorb a major share of macro demand. At least one AI lab may become a $10TN company within 5 years. 2. The Capacity Bottleneck: Demand Doubling Overnight For top infrastructure and data providers, growth is no longer limited by customer acquisition. It is limited by execution. Demand is scaling so fast that leading companies could double revenue overnight if they had enough capacity. The challenge now is how quickly they can mobilize specialized human networks and build high-fidelity environments for enterprise demand. 3. Is the Stated Revenue Really Revenue or GMV? The stated revenue is not GMV because the talent network is only one part of a vertically integrated value chain. Customers buy complete tasks for model improvement, not simple marketplace listings. With 30% to 40% gross margins, the business owns the full lifecycle, from sourcing experts to deploying AI project managers and running quality checks. 4. The Inversion of Corporate Opex: Token Spend vs. Salaries In high-growth AI companies, token spend for internal agents has already surpassed employee headcount costs. As operations, interviewing, accounting, and fraud detection move to agents, capital allocation shifts from salaries to inference compute. 5. Why Token Spend Inside Companies Will Keep Increasing Driven by Jevons Paradox, enterprise token consumption will keep rising as models improve and costs fall. Companies will use more compute to unlock higher-order reasoning, not less. F500s are responding by building evaluation systems that let them hot-swap models and optimize inference budgets. 6. The Tens of Millions Talent War for AI Researchers The market for top AI researchers is severely supply constrained, with demand far above available talent. Companies are offering compensation packages worth tens of millions in stock per year to secure elite researchers. This wage spike shows that world-class research talent remains the core bottleneck in AI. (links in comments)

We’re running a 24-hour hackathon on June 19–20 in San Francisco in partnership with Cognition, Etched, and Anthropic. There’s three tracks to build in: - Agents: Multi-step systems that take a goal and execute across tools, plans, and long horizons. - Real-Time and Interactive: Systems that respond in the moment, across modalities. - Talent Marketplace + Applied AI: Systems that measure what people and models can actually do. Autograders, rubrics, skill verification, and human-in-the-loop evaluation. Our guest judges include: Brendan Foody, Robert W. and Steven Hao. We’re offering a $50k top prize with $100k+ in total awards. Every accepted team gets 8xH100s, Cognition API access, and Anthropic credits. Come build with us. Apply by 6/12 at the link in the comments.