Mercor
Mercor matches AI labs with vetted domain experts and publishes APEX benchmarks for evaluating frontier models on real professional work.
Pick Mercor when the bottleneck is credentialed human judgment — a CPA stress-testing an agent's book close, a lawyer grading a redline — or when you want benchmarks authored by people who publish their methodology. The $103/hr average rate tells you the economics before you scope anything. If you need high-volume cheap labeling, this is the wrong vendor.
Verified 1d ago · liveness 82/100 · cite: rightaichoice.com/tools/mercor
- Frontier AI labs needing credentialed specialists to build or grade professional-domain evals
- Enterprises that want to monetize workflow data through anonymization while retaining control
- Physicians, lawyers, CPAs and engineers seeking project-based AI work at $60-$77/hr listings with daily payouts
- Teams that already cite APEX scores and want the benchmark authors on the evaluation side
- Buyers who need high-volume, low-cost data annotation rather than credentialed specialist judgment
- Teams wanting a fully managed evaluation pipeline without dedicating internal staff to run it
- Organizations with strict security review requirements that have not yet read Mercor's June 2026 incident disclosure
We scan live Reddit threads, YouTube comments, X posts, G2 reviews and other communities — and hand you an honest verdict in under a minute.
- Honest verdict, not marketing
- Real pros & cons from real users
- Attributed quotes with receipts
3 free scans · no card needed
Skip Mercor if you need cents-per-label volume annotation from a crowd, or if you want a turnkey eval pipeline you can run without assigning any internal staff to define the tasks and grade against them.
Budget against expert time at roughly $61-$150/hr contracted rather than a flat software seat fee. For comparison, Surge AI and Scale AI sell comparable specialist data with more published rate cards, so push for a per-task and per-hour breakdown early.
In short
Mercor — Mercor matches AI labs with vetted domain experts and publishes APEX benchmarks for evaluating frontier models on real professional work. Best for Frontier AI labs needing credentialed specialists to build or grade professional-domain evals, Enterprises that want to monetize workflow data through anonymization while retaining control, Physicians, lawyers, CPAs and engineers seeking project-based AI work at $60-$77/hr listings with daily payouts. Contact Sales pricing.
No head-to-head comparison for Mercor yet — the alternatives below are the closest matches we list.
What's new in Mercor
Checked 3 days agoAcross the latest 5 updates: 2 feature updates, 1 changelog entry and 2 news mentions.
Mercor is building a research team
Mercor announced a dedicated research team focused on measuring AI on real work, building expert data around where models fail, and post-training to close the gap.
Human Baselines for Benchmarks: AI Now Outperforms Junior Accountants
Mercor tested 12 licensed CPAs against frontier AI on APEX-Accounting tasks and reported that AI was faster, more accurate and 10x cheaper.
How Mercor 20x'd LLM eval throughput
An engineering post detailing how Mercor increased LLM evaluation throughput by 20x.
Introducing APEX-Agents 1.1
Mercor released APEX-Agents 1.1, an update to its benchmark for long-horizon, cross-application agent tasks in professional services.
SWE-Marathon-Ext: Can AI build a whole SaaS product?
An engineering post on SWE-Marathon-Ext, an extended evaluation testing whether AI can build a complete SaaS product.
What people actually say about Mercor — is it worth it?
We ran a structured research pass across product reviews, community discussions, and post-purchase forum threads to surface the patterns vendors won't publish themselves. Below: the recurring strengths, the hidden costs people mention most, and the cohort that consistently regrets adopting this tool.
5 mentions across 4 sources (Hacker News, YouTube, Product Hunt, Lemmy), 56 more we could not attribute · researched Oct 8, 2026.
Weighted by the 61 posts each of 4 sources contributed.
- +Average contracted expert rate of $112/hr with listed roles paying $75–$120/hr
- +One expert reported $1k in 12.5 hours at roughly $80/hr effective
- +Daily payouts instead of 30- or 60-day freelance invoice cycles
- +447k+ roles created and $4M+ in reported daily payouts signal real scale
- +APEX-Agents, APEX-SWE with Cognition, and APEX-Accounting with Ramp benchmarks
- −Onboarding takes hours and accounts can stay unactivated for up to 7 days
- −At least one honest reviewer said 'overall summary — don't waste your time'
- −4TB breach affecting 40k contractors led Meta to pause work with Mercor
- −No public pricing page — every enterprise budget talk requires a sales demo
- −Task monotony and screen-time fatigue recur across YouTube expert reviews
- • No published pricing — budget discovery requires a sales demo cycle
- • Setup and waiting time before accounts activate (up to 7 days unpaid)
- • Enterprise data-security review and remediation costs post-breach
- • Potential legal or compliance review when selling workflow data
Viability Score
How well maintained and how widely used is Mercor? Built from what the vendor actually publishes (docs, changelog, tutorials, integrations, pricing), whether the site is live, and how much real users discuss it. How we calculate this
Last calculated: October 2026
How we score →Key Features
- Expert marketplace matching AI labs with vetted physicians, lawyers, CPAs and engineers
- Average contracted expert rate of $103/hr across the network
- 578k roles created and $4M+ in daily payouts to experts
- Daily expert payouts instead of 30- or 60-day freelance cycles
- APEX-Agents benchmark for long-horizon cross-application professional services tasks (Gemini 4 Argon High 82.4%)
- APEX-1 single-turn benchmark across investment banking, corporate law, management consulting and medicine
- APEX-SWE software engineering benchmark built with Cognition (Opus 5.5 Max 67.6%)
- APEX-Accounting benchmark built with Ramp (Gemini 4 Argon High 71.5%)
- Legal benchmark of 1,749 attorney-built tasks across 24 practice areas, built with Harvey
- τ³-Banking fintech customer-support benchmark built with Sierra AI over 700 policy documents
- SWE-Marathon-Ext extended evaluation testing whether AI can build a whole SaaS product
- APEX-Agents 1.1 update to long-horizon cross-application agent evaluation
- Human data and evals drawn from 30k+ domain experts
- RL environment construction built on real professional work rather than synthetic tasks
- Data monetization with automated anonymization masking 60+ categories of sensitive identifiers
About Mercor
Mercor runs a two-sided operation for frontier AI development. On one side, AI labs and enterprises need credentialed humans — physicians, lawyers, CPAs, engineers, consultants — to build training data, write evals, and grade model output on tasks where being wrong costs money. On the other, over 30,000 specialists pick up project work at an average contracted rate of $103/hr, paid daily rather than on the usual 30-to-60-day freelance lag. The platform reports 578k roles created and $4M+ in daily payouts. The research arm is what separates Mercor from a staffing marketplace. Its APEX benchmark family is published openly with leaderboards, papers, and code: APEX-Agents for multi-hour professional-services tasks across investment banking, corporate law and management consulting (Gemini 4 Argon High leads at 82.4%), APEX-1 for single-turn professional domains (Gemini 4 Argon High, 81.1%), plus vertical benchmarks built with partners — APEX-SWE with Cognition, APEX-Accounting with Ramp, a 1,749-task legal benchmark with Harvey, and the τ³-Banking fintech benchmark from Sierra AI. The October 2026 announcement of a dedicated Mercor Research team formalizes the shift: measure AI on real work, build expert data around where models fail, then post-train to close the gap. Supporting work is public too, including SWE-Marathon-Ext testing whether AI can build a complete SaaS product, and an October 2026 study running 12 licensed CPAs against frontier models on APEX-Accounting tasks. For enterprise buyers the offering splits three ways: human data and evals from the expert network, data monetization that anonymizes 60+ categories of sensitive identifiers so a company can sell workflow data while keeping control, and enterprise AI deployment with embedded Forward Deployed Engineers. Positioning-wise, Mercor sells credentialed judgment and benchmark credibility, not cheap annotation volume — buyers wanting per-unit labeling economics should look at Surge AI or Scale AI's
Behind the Verdict
The honest way to read Mercor is as two products sharing one brand. There's the expert marketplace, and there's the research org that ships APEX benchmarks with leaderboards and code. Which one you're buying depends on whether your problem is 'we need 40 lawyers to grade 2,000 memos' or 'we need someone to tell us where our agent actually fails.' We'd reach for it when credentialing is the point. A clinician reviewing clinical reasoning, a CPA checking a close, an engineer auditing a bug fix — those tasks punish crowd workers and reward domain experts, and Mercor's network is built for exactly that. The daily payout structure also matters more than it sounds: it's how you keep good specialists coming back for the second and third project. The APEX work is worth citing on its own. Vertical benchmarks built with Cognition, Ramp, Harvey and Sierra AI are narrow but unusually concrete, and the October 2026 CPA study — 12 licensed accountants against frontier models on accounting tasks, with AI reported faster, more accurate and roughly 10x cheaper — is the kind of finding most vendors keep private. A research team that publishes where models fail is a useful counterparty for eval work. Where it bites: scope comes before price. There's no published per-task rate card, so budget planning starts with a conversation, not a checkout page. Enterprise AI engagements lean on Forward Deployed Engineers, which means your team has to be in the room — this isn't a drop-in managed pipeline. Security-conscious buyers should read Mercor's June 2026 incident disclosure and its security page before the procurement call, not after. And if your workload is millions of simple labels, the $103/hr average makes the math obvious: go to Surge AI or Scale AI's crowd tiers and keep Mercor for
Researching Mercor? Get your full AI stack in 60 seconds.
Free, no signup — tell us your goal and get tools matched to your budget & existing stack.
Real-world workflow fit
Concrete scenarios for the personas Mercor actually fits — and what changes day-one when you adopt it.
You need credentialed graders for a professional-services agent before launch. You scope the task with Mercor, they recruit and vet physicians or consultants matching the domain, and the experts grade model outputs against your rubric using real professional workflows rather than synthetic prompts.
Outcome: A graded eval set built on real professional work, with results you can compare against published APEX scores instead of a one-off internal rubric.
Your firm generates workflow data it wants to license without exposing client records. Mercor's anonymization layer masks 60+ categories of sensitive identifiers and your organization controls what is shared, so the data can be packaged for an AI lab buyer.
Outcome: A new revenue stream from existing workflow data with identifier masking applied before anything leaves your control.
You have AI pilots stuck in proof-of-concept. Mercor maps your workflows and institutional knowledge to find the high-value opportunities, then embeds Forward Deployed Engineers to move the work into production and measure results.
Outcome: Production AI deployed against identified high-value workflows, with evaluation built in to track performance after launch.
Use Cases
- Label legal documents for AI document understanding systems.
- Provide expert feedback on medical AI to improve clinical decision support.
- Evaluate model-generated equity research reports for accuracy.
- Generate high-quality training data for AI agents in software engineering.
- Assess model performance on professional tasks like consulting analysis.
- Staff clinicians for ambulatory and community care insight studies at $90/hr.
- Recruit financial crime, AML and KYC analysts for model evaluation work.
- Benchmark frontier models against licensed CPAs on accounting close tasks.
Models Under the Hood
as of 2026-10-22
Limitations
- Mercor is a human-expertise marketplace and AI evaluation service rather than an AI product; the model names on the site (e.g.
- Gemini 4 Argon High, Opus 5.5 Max, Fable 5.1 Max, GPT-6 Astra Max) appear only as benchmark leaderboard subjects across APEX benchmarks.
- The expert network is gated — experts must apply and be vetted, and available roles depend on demand from AI labs.
- The company publicly disclosed a security incident in June 2026.
- Enterprise and data-monetization engagements run through contact-sales channels.
as of 2026-10-10
Verification history
We have re-verified Mercor 21 times since . Each pass re-reads the vendor's own pages and re-checks every listed field against that evidence; passes where nothing had changed are marked as such.
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
Showing the 6 most recent of 21 verification passes.
Free to cite with attribution — this page re-verifies continuously.
Where the pricing makes sense
The company stage and team size where Mercor's pricing actually pencils out — and where peers do it cheaper.
Budget against expert time at roughly $61-$150/hr contracted rather than a flat software seat fee. For comparison, Surge AI and Scale AI sell comparable specialist data with more published rate cards, so push for a per-task and per-hour breakdown early.
Setup time & first value
How long it actually takes to get something useful out of Mercor — broken out by persona, not the marketing-page minute.
For experts, applying and vetting takes days to a few weeks depending on the role; payouts then settle daily. For enterprise buyers, expect a scoping and sales cycle before work starts, then a ramp while Mercor recruits and vets the specific specialists for your tasks. Labs that already have a defined rubric move faster than buyers still deciding what to measure.
Switching to or from Mercor
How to bring data in from common predecessors and how to get it back out — written for the switcher, not the buyer.
- →From Upwork or general freelance marketplaces: bring your existing task rubric to Mercor and have credentialed specialists recruited and vetted against it.
- →From Scale AI or Surge AI crowd tiers: layer specialist APEX-style evaluation on top of any existing volume-labeling pipeline.
- →From in-house contractor benches: replace ad-hoc clinician, lawyer or CPA contractors with a marketplace that handles vetting and daily payouts.
- ↗To Surge AI or Scale AI: if your need shifts to high-volume, lower-cost annotation rather than credentialed specialist judgment.
- ↗To a pure staffing marketplace: if you want to hire experts directly and manage the project yourself without an attached eval methodology.
Integrations
Resources & Guides
Tutorials & Learning
YouTube returned 6 videos for “Mercor”, and we withheld 6: 6 could not be judged, because “Mercor” is a single word that other videos use for other things. We are showing none, because we could not prove any of them are about Mercor.
Official links
Tools that pair well with Mercor
Common stack mates teams adopt alongside Mercor, with the specific reason each pairing earns its keep.
Labelbox
Labelbox builds RL environments frontier labs train on and background agents engineering teams hand their backlog to.
Outlier AI
Outlier AI pays subject-matter experts to train AI models remotely through flexible, project-based prompt writing and answer grading.
AfterQuery
AfterQuery turns expert reasoning into SFT, RL rubric, agent, and computer-use training data for frontier models.
Alternatives to Mercor
View all
Labelbox
Labelbox builds RL environments frontier labs train on and background agents engineering teams hand their backlog to.
Outlier AI
Outlier AI pays subject-matter experts to train AI models remotely through flexible, project-based prompt writing and answer grading.
AfterQuery
AfterQuery turns expert reasoning into SFT, RL rubric, agent, and computer-use training data for frontier models.
Frequently Asked Questions
Categories
Best-of guides
Topics
Used Mercor? Help shape our editorial sentiment research.