What Mercor does

Mercor, founded in 2023 and headquartered in San Francisco, operates an AI-powered platform with two principal offerings: a curated marketplace connecting 30,000+ vetted domain experts with AI labs for model training, evaluation, and human feedback, and an enterprise services arm focused on custom AI agent deployment. The platform spans 300+ professional domains including law, medicine, finance, consulting, and software engineering, with experts paid an average contracted rate of $80/hour and the marketplace processing more than $3 million in daily payouts. Mercor has developed the APEX family of benchmarks (APEX v1, APEX-Agents, APEX-SWE, and ACE), which measure frontier AI model performance on economically valuable professional and consumer tasks, and publishes research on arXiv establishing thought leadership in AI evaluation methodology.

Mercor's technology stack includes proprietary reinforcement learning environments, expert-created training datasets spanning 50,000+ tasks across 30+ domains, enterprise data anonymization pipelines featuring 60+ identifier detection types, and infrastructure for orchestrating high-volume global expert engagements. The company has built AI-assisted interviewing systems (more than 5 million interviews conducted), AI-driven assessment and matching engines, and tools for deploying custom AI agents within enterprise workflows, with disclosed customer outcomes including 90% of support requests resolved end-to-end by AI and a 73% reduction in ticket close time.

The company monetizes through transaction fees on expert engagements with frontier AI labs (OpenAI, Anthropic, Meta, and Google), managed services for enterprise AI deployments, licensing of pre-built training datasets and benchmarks, and data partnerships with enterprises that contribute anonymized operational data to AI labs. Revenue scaled from approximately $1 million to $500 million annualized run rate within roughly 16 months (early 2024 to late 2025), with the company referencing a $1 billion+ annualized run rate in 2026. Mercor reached a $10 billion post-money valuation in October 2025 following a $350 million Series C led by Felicis.

Name

Mercor

Legal name

Mercor.io Corp

Website

https://mercor.com

Company type

Private

Founded year

2023

Operating status

Operating

Headcount range

5,001–10,000 employees

Short description

Mercor operates an AI-powered marketplace connecting 30,000+ vetted domain experts with frontier AI labs for model training, evaluation, and custom enterprise AI agent deployment, supported by proprietary APEX benchmarks and a $10B valuation.

Ownership category

akta.pro rank

Product category

AI Training Data and Evaluation Platform

NAICS

Computing Infrastructure Providers, Data Processing, Web Hosting, and Related Services (518210), Software Publishers (513210), Custom Computer Programming Services (541511)

SIC

Services-Computer Processing & Data Preparation (7374)

akta.pro primary industry

Data, Analytics & AI Training (BI, Data Engineering, ML/GenAI) (BPAEANAH)

akta.pro secondary industries

AI Observability, Monitoring & Evaluation Platforms (Drift, Quality, Safety) (HDAEANAF), Industry/Domain Analytics Solutions & Accelerators (BPAEAHAK)

Keywords

• AI talent marketplace
• AI training data
• RLHF services
• AI benchmarking
• Enterprise AI agents

Headquarters

HQ city

San Francisco

HQ country

United States

HQ region

North America

Offices3 records

Markets served

GTM type

B2B

Offering type

Software

Cost components

Personnel, Technology or R&D, Marketing or Sales, Operations, Infrastructure

Revenue model

1. AI Training and Evaluation Services: Mercor connects AI labs (OpenAI, Anthropic, Meta) with domain expert contractors who train and evaluate AI models. The company generates revenue by facilitating these engagements, paying contractors from the fees charged to AI lab clients. Annualized revenue grew from $100M to $500M run rate between early 2024 and late 2024.
2. Enterprise AI Agent Deployment: Mercor Enterprise offers custom AI agent deployment, expert staffing, and data monetization services to Fortune 500 companies deploying AI in production. The company helps enterprises encode their own knowledge into AI systems and benchmark AI product effectiveness.
3. Enterprise Data Licensing and Monetization: Mercor partners with enterprises to anonymize and contribute operational data (workflow, messaging, code, finance data) to top AI labs, receiving compensation for the data. Enterprises get paid and gain insights into AI transformation.
4. Benchmark and Dataset Licensing: Mercor licenses pre-built AI training datasets (APEX family, ACE, professional task datasets) to AI labs and enterprises, enabling immediate access to expert-created training data without custom pipeline development.

Pricing tiers

Go-to-market motion2 records

Distribution channels3 records

Marketing channels5 records

Core offering

Mercor operates a two-sided AI talent marketplace and platform that connects 30,000+ vetted domain experts (doctors, lawyers, engineers, consultants) with frontier AI labs for model training, RLHF, and evaluation. The company also develops the APEX family of AI benchmarks, licenses off-the-shelf training datasets (50K+ tasks, 1M+ expert annotations across 30+ domains), and deploys custom AI agents for Fortune 500 enterprises through Mercor Enterprise and Mercor Data offerings.

Product overview

Mercor is organized around two core offerings: (1) an AI benchmarking and evaluation platform featuring the APEX family of benchmarks (including APEX, APEX-Agents, APEX-SWE, and ACE) that measure AI model capabilities on professional tasks, and (2) a talent marketplace connecting domain experts with AI labs for model training and evaluation. The company also offers Mercor Enterprise for custom AI agent deployment and Mercor Data for enterprise workflow data partnerships. Founded in 2023, the company has scaled to $10 billion valuation with over $1B+ revenue run rate.

Differentiator

Problem solved

Functional benefit

Products and services

• Mercor Experts (Talent Marketplace) Two-sided marketplace that connects domain experts (doctors, lawyers, engineers, consultants, finance professionals) with AI companies for RLHF, evaluation, and task-based data creation. Professionals self-enroll through work.mercor.com and are matched to AI training projects with average contracted rate of $80/hr.
• Mercor Enterprise Custom AI agent deployment service for enterprises that builds AI agents trained on the customer's specific workflows, standards, and organizational knowledge. Includes agent diagnostics, deployment, benchmarking, and integration services.
• Mercor Data Enterprise data partnership program that allows companies to monetize anonymized operational data (messaging, documents, email, CRM, code, finance data) by contributing it to top AI labs for model training. Includes anonymization pipeline with 60+ identifier detection types and AWS KMS encryption.
• APEX Benchmark Family Standardized AI evaluation benchmarks that measure frontier model capabilities on economically valuable professional tasks. Includes APEX v1 (professional services), APEX-Agents (long-horizon agent tasks), APEX-SWE (software engineering), and ACE (consumer tasks). Published with leaderboards to drive industry adoption.
• Off-the-Shelf Data License-ready AI training datasets covering professional services, coding, research, agentic, and multimodal tasks. Includes APEX Agents (2,620 tasks across 8 domains), ACE (1,200 tasks across 4 consumer domains), APEX v1 (920 tasks across 5 professional domains), and BrowseComp (3,558 web-browsing tasks). Sample tasks delivered same day.

Quantifiable outcome

• Revenue grew from $1M to over $500M annualized run rate in under 16 months (2024)
• +5 more outcomes

Named customers4 records

Segments3 records

Ideal customer profiles2 records

Technology focussed Yes

API detail

Has API

No

API docs

API detail

Core technology

AI maturity

App detail

Integration1 record

AI capability9 records

Feature4 records

Partnerships

Six partnerships are on record, tiered core infrastructure, research partner, core customer, former core customer and customer.

• Stripecore infrastructureTechnology or Integration · 1 April 2026Mercor uses Stripe Connect to power its marketplace operations, including onboarding professionals, processing global payouts, and maintaining cross-regional compliance. Stripe Instant Payouts are used for faster contractor earnings access.
• Cognitionresearch partnerStrategic or Co-development Partner · 24 March 2026Mercor collaborated with Cognition to create the APEX-SWE benchmark, which measures whether frontier AI models can handle real software engineering work. Cognition reviewed integration and observability tasks created by Mercor's platform engineers.
• OpenAIcore customerTechnology or IntegrationOpenAI is a major customer of Mercor, sourcing human expert feedback and AI training data through Mercor's contractor network. OpenAI uses Mercor to train and evaluate frontier AI models.
• Anthropiccore customerTechnology or IntegrationAnthropic sources AI training and evaluation services from Mercor's expert network, utilizing domain experts to improve model performance on complex professional tasks.
• Metaformer core customerTechnology or IntegrationMeta was a major client of Mercor for AI training data services. Meta suspended all work with Mercor in April 2026 following a data breach, representing a significant business loss for the startup.
• GooglecustomerTechnology or IntegrationGoogle is listed as a client using Mercor's expert network for AI model training and evaluation.

Scale indicators8 records

Recent moves6 records

Expansion highlights7 records

Direct peers

• Scale AI: Scale AI is the dominant provider of training data, RLHF, and evaluation services for frontier AI labs including OpenAI, Anthropic, Meta, and others. It is the closest direct competitor to Mercor's expert network and AI training/evaluation offerings, with overlapping customers and comparable use cases.
• Surge AI: Surge AI operates a high-quality data labeling and RLHF platform serving AI labs, with deep bench of expert annotators for frontier model training. It is closely comparable to Mercor's expert marketplace and evaluation offerings, competing head-to-head for AI lab training budgets.
• Outlier.ai: Outlier connects domain experts (mathematicians, scientists, writers, coders) with AI companies for RLHF, evaluation, and reasoning task work. It targets the same expert-supplier model as Mercor's talent marketplace, serving the same buyer segment (frontier AI labs).
• Turing: Turing operates an AI-vetted developer and expert talent marketplace that also pivoted toward RLHF services for frontier AI labs. Its talent matching and AI-training data services overlap with Mercor's expert network, with comparable end buyers.
• Invisible: Invisible Technologies operates AI training data operations including expert labeling and RLHF workflows for frontier AI labs. It is a direct competitor in the AI training data/expert operations space, comparable to Mercor's enterprise services arm.

Broad incumbents

• Appen: Appen is a long-standing public data annotation and crowd-sourcing company serving ML/AI teams globally. It is broadly comparable to Mercor's training data business but operates at a larger, more diversified scale with global human-intelligence infrastructure rather than expert-only positioning.
• Toptal: Toptal runs an elite freelance network (top 3% of talent) offering software engineers, designers, and finance/legal experts on demand. It is comparable to Mercor's expert marketplace in business model and curated talent positioning, though not focused exclusively on AI training.
• Upwork: Upwork is a public, large-scale freelance marketplace spanning developers, writers, designers, and consultants. It is adjacent to Mercor's marketplace model and overlap with the AI-training workflow use case, though significantly broader and lower-priced on average.

Emerging players

• Labelbox: Labelbox provides data labeling and evaluation platform infrastructure for AI/ML teams. It is comparable to Mercor's APEX evaluation offering in the model assessment workflow, though more focused on tooling than expert sourcing.
• Clickworker: Clickworker provides crowd-based data annotation and AI training data generation across global markets. Comparable to Mercor in the AI training data category but at lower price points and broader crowd sourcing rather than vetted domain experts.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat5 records

Key risks6 records

Key highlights7 records

Customer concentration

• LinkedInlinkedin.com/company/mercor-ai
• Xx.com/mercor_ai
• Facebookfacebook.com/MercorSoftware

Compliance2 records

Revenue estimate

Valuation estimate

Number of profiles

Profiles4 records

Funding overview

Funding rounds4 records

Investors13 records

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A2 records

Investments1 record

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about Mercor