How much does Outlier actually pay?
Outlier is one of the largest AI training platforms, hiring across coding evaluation, math reasoning, expert domain data, and generalist annotation. This page tracks Outlier's disclosed hourly rates by task type — sourced from listings on HireFeed, not surveys.
Outlier pay rates by task type
Outlier's roles span a wide pay range because task type and required expertise vary sharply. Here's the current picture from Outlier-tagged listings on the HireFeed feed. Full methodology in the Pay Index.
| Task type | Typical range | Top decile | What it involves |
|---|---|---|---|
| Math Reasoning · PhD | $60–$120/hr | $150/hr | Proof verification, multi-step reasoning eval, olympiad-tier problems |
| Coding Evaluation | $40–$75/hr | $95/hr | Side-by-side code review, debugging, correctness rating |
| Creative Writing Trainer | $30–$50/hr | $65/hr | Fiction, non-fiction, tone calibration for language models |
| Multilingual QA | $25–$45/hr | $60/hr | Translation quality, localization edge cases, 40+ languages |
| Generalist Annotation | $18–$25/hr | $32/hr | Preference ranking, basic QA, simple task grading |
| Domain Expert · Specialist | $80–$180/hr | $220/hr | Law, finance, engineering — credential-verified projects |
Median of Outlier-tagged pay-disclosed listings, trailing 30 days. Illustrative during preview.
Why does Outlier pay vary so much?
Outlier operates as an AI training marketplace across many task categories at different levels of expertise required. Rate variance is driven by three things:
- Task complexity. Reviewing a single-turn preference pair is different from evaluating a multi-step agentic trajectory. Higher cognitive load → higher rate.
- Domain credential required. Math PhDs, medical MDs, and JDs command 2-4x the generalist rate because the pool is smaller and credentials are verified.
- Worker tier. Outlier uses a Generalist → Specialist → Expert ladder. Higher tiers earn more per task but require passing calibration assessments.
How does Outlier's pay compare to other platforms?
Broadly: Outlier is competitive on coding and math roles, mid-market on generalist annotation, and slightly below Mercor/Surge on credentialed domain expert work. See the full Pay Index for cross-platform comparison. Note that pay bands shift week-to-week as labs launch new projects.
Reality check: Pay is only one variable. Worker complaints on Outlier center on queue reliability (roles going dark with zero notice) and inconsistent screening feedback — factors this page can't quantify. Check community reports before committing time to a long screening funnel.
Common questions.
How much does Outlier pay per hour in 2026?
Outlier's disclosed hourly rates range from $18-$25/hr for generalist annotation to $60-$120/hr for PhD-level math reasoning and $80-$180/hr for domain experts (law, medicine, specialized science). Median coding evaluation rate is around $50-60/hr. Top decile across all Outlier roles is roughly $220/hr for credentialed domain expert work.
Does Outlier pay weekly or monthly?
Outlier pays weekly for approved work, via direct deposit or PayPal depending on region. Payment is tied to task completion and quality review, so there's typically a 1-2 week lag between work completed and payment received.
Is Outlier legit?
Outlier is an established platform operated by Scale AI's contractor arm. It pays real money for real AI training work at rates disclosed in listings. That said, worker experience varies — the most common complaints are queue instability (work drying up without warning), long screening funnels, and inconsistent quality feedback. See our trust checklist framework for how to evaluate any AI training platform.
Do you need a PhD to work on Outlier?
No, though credential level determines eligibility for specific projects. Outlier hires across a spectrum from generalist annotation (no formal credential required) to PhD-only math and science reasoning roles. Higher-rate projects gate on credential verification.
How long is the Outlier screening process?
Screening typically involves a domain-specific assessment (30-90 minutes) plus a short calibration task set. Passing takes anywhere from 1 day to 2 weeks depending on the project's queue depth and reviewer availability.
2,000+ rate-disclosed jobs, added today.
Every listing shows pay upfront. Apply direct. No middleman.