Freelancing & Fiverr
Mercor vs Tortoise vs Outlier: AI evaluation platforms compared
Three growing AI evaluation platforms compared on acceptance bar, hourly rate and task supply.
Published
The AI evaluation market has expanded beyond Mercor and Scale in 2026. Three platforms account for most expert paid evaluation work: Mercor, Tortoise, and Outlier (Scale AI brand).
Mercor
High vetting bar. Rates typically 60 to 100+ USD/h for software and technical tracks. Long-form tasks with depth. Best for experienced specialists who value rate over volume.
Tortoise
Newer entrant focused on premium tracks (formal reasoning, advanced math, scientific evaluation). Smaller pool of work but high acceptance rate for genuinely scarce profiles. Rates 50 to 150 USD/h for premium tracks.
Outlier (Scale AI)
Larger volume, lower acceptance bar. Rates typically 18 to 45 USD/h. Steady weekly volume. Best for evaluators wanting predictable hours over rate.
Side by side
| Factor | Mercor | Tortoise | Outlier |
|---|---|---|---|
| Typical rate (dev) | 60-100+ USD/h | 50-150 USD/h | 18-45 USD/h |
| Acceptance bar | High | Very high (niche-dependent) | Low-medium |
| Volume reliability | Bursty | Bursty | Steady |
| Task complexity | Long-form | Premium-niche | Mixed |
| Best for | Specialists | Scarce-domain experts | Volume seekers |
Stacking
Many evaluators run two or three platforms in parallel. Mercor and Tortoise for rate, Outlier for filling gaps. Exclusivity is rare; check before signing.
What is changing in 2026
Premium domains (formal verification, advanced math, specific legal) command higher rates everywhere as AI labs push reasoning capabilities. Generic code-review tracks have softened. The gap between top and entry rate is widening.
Sister site · ai-by-job.com
Best AI tools for tech work
The best AI tools for every job, in English, French and Spanish.
Explore on ai-by-job.com →Earn with AI (or build with it)
Affiliate · disclosure belowTwo options if AI is in the loop: get paid by Mercor to evaluate model outputs, or work alongside Claude on your own product.
Get paid to evaluate AI model outputs. Vetted experts typically earn $60-$100+/hour reviewing prompts, code, and reasoning.
Apply to Mercor →
The LLM we use to research and draft our guides. Strong at long documents, code and careful reasoning.
Try Claude →
We may earn a small commission if you sign up via these links. Pricing and features are unchanged.
Freelance marketplace
Turn your skills into income on Fiverr
Fiverr is the fastest way to start selling tech services to buyers worldwide. List a gig in minutes, set your own prices, and get paid for code, automation, design and AI work.
Start selling on Fiverr →Get paid by AI labs
Earn $60-$100+/hour evaluating AI model outputs
Mercor matches vetted experts (developers, researchers, domain specialists) with paid evaluation work for frontier AI labs. Async, remote, USD payouts. Best fit if you have technical depth and want flexible high-rate side income.
Apply to Mercor →Frequently asked questions
Which platform pays AI evaluators the most in 2026?
Tortoise leads on premium niches; Mercor leads for mainstream technical tracks. Outlier pays less per hour but offers more sustained volume.
Can I work on multiple AI evaluation platforms at once?
Yes. Exclusivity is rare. Most experienced evaluators stack at least two platforms to balance rate and volume.