Outlier, Mercor, and Surge AI dominate the conversation about AI training contracts. But there are at least eight other platforms paying competitive rates with smaller, less-saturated applicant pools. Here's the short list.
1. Invisible Technologies
Originally a "human-AI hybrid" ops automation company, Invisible has grown into a serious frontier-lab partner running large RLHF and red-teaming operations.
- Pay: $30–$70/hr generalist; $90–$140/hr specialty.
- Hours: Highly variable — 8–25 hrs/wk depending on active programs.
- Strength: Long-running structured programs that pay even during downtime weeks. Less feast-or-famine than gig platforms.
- Watch out: Their application is rigorous. Expect 2–3 written rounds.
2. DataAnnotation.tech
Smaller, more straightforward; consistent pay; less pressure than Outlier on hours.
- Pay: $20–$45/hr generalist; up to $65/hr for tested specialties.
- Hours: 6–18 hrs/wk; consistent.
- Strength: Easier application, decent training, friendly interface.
- Watch out: Lower ceiling on rates than Outlier or Mercor.
3. Remotasks
Now part of Outlier's parent (the Scale → Outlier consolidation), but Remotasks still operates as a separate brand for specific task types — particularly image annotation, lidar, and self-driving training data.
- Pay: $15–$35/hr for image/video annotation; $40–$80/hr for high-precision tasks.
- Hours: 10–25 hrs/wk; depends on active campaigns.
- Strength: Different task type than text-based platforms — good if visual-spatial work suits you.
- Watch out: Lower text-task rates than the major platforms.
4. Prolific
Academic-research-focused. Pays per task at calculated rates; transparent and well-regulated.
- Pay: Effective rate of £8–£15/hr ($10–$19) for surveys; can hit $25–$40/hr for specialized research panels.
- Hours: 4–10 hrs/wk; very part-time.
- Strength: Lowest barrier to entry of any platform on this list. Quick onboarding, fast pay.
- Watch out: Hourly rate ceiling is lower; treat as supplementary, not primary.
5. Toloka
Russian-origin, now globally distributed. Strong for non-English language work.
- Pay: $5–$25/hr generalist; $30–$50/hr for less-resourced languages.
- Hours: Highly flexible; can be 20+ hrs/wk if you target the right tasks.
- Strength: Multilingual depth no other platform matches. Native speakers of less-supported languages can do well here.
- Watch out: Generalist rates are low; specialty rates are reasonable but not exceptional.
6. CloudFactory
Africa- and Nepal-headquartered. Different economics — strong for contractors in those regions.
- Pay: $4–$12/hr base; $15–$30/hr for specialty (legal, medical, financial).
- Hours: 20–35 hrs/wk; designed for primary income in lower-cost-of-living markets.
- Strength: Stable, full-time-ish work; predictable schedules.
- Watch out: Base rates are below US/EU norms — designed for cost-of-living arbitrage, not maximum income.
7. Labelbox (now Encord-merged)
Primarily a tooling company, but they run their own labeling workforce on top of their platform for premium customers.
- Pay: $25–$60/hr generalist; $80–$120/hr specialty (medical imaging especially).
- Hours: 8–20 hrs/wk; sporadic but well-paid.
- Strength: Top-tier tooling. The work is more like "annotate using a real product" than "fight a clunky web form."
- Watch out: Pool is small; selective application process.
8. Vector AI / Domain-specific labs
This isn't one platform — it's a category. A growing number of vertical AI startups (especially in legal, medical, and finance) hire their own labelers directly. Examples in 2026: Harvey (legal AI), Hippocratic (medical AI), Magic (coding AI), various proprietary fintech model trainers.
- Pay: $80–$200/hr; the highest rates on this list, by far.
- Hours: 4–12 hrs/wk; project-based.
- Strength: Top of the rate range. Often direct relationships with the labs that need the work.
- Watch out: Hardest to find. They don't post on standard platforms — you have to follow the labs themselves and apply when they open contractor slots. (joblet.ai surfaces these when they're open.)
How to play this
The optimal strategy in 2026 is not to pick one platform — it's to layer:
- Primary platform for hours: Outlier or Mercor.
- Specialty platform for high rate: Vector AI / domain-specific lab in your specialty if applicable.
- Backup for consistency: DataAnnotation or Invisible.
- Multilingual gigs (if you qualify): Toloka or Prolific.
Treat it like a small portfolio. Most successful contractors we track are active on 2–4 platforms simultaneously, optimizing across them weekly based on which has the most attractive task pool.
Application order
Apply to all of them at once. The bottleneck on this work isn't capacity — it's the application backlog. If you wait to be approved at Outlier before applying to Mercor, you've wasted 3–6 weeks. Stack the applications and pick the offers as they come in.
Bottom line
Outlier and Mercor get all the attention. The other six platforms on this list collectively account for ~30% of the AI training market in 2026, with less competition for slots and surprisingly competitive specialty rates. If you only know about the top 2 names, you're leaving money on the table.