Remotasks was historically the entry-tier AI training platform — easy to start, predictable pay, broad task variety. Then Scale AI's brand became Outlier, and Remotasks's place in the ecosystem shifted. Here's what Remotasks looks like in 2026 and whether contractors should still apply.
What changed
Remotasks now operates as a sub-brand under the Outlier parent. The change isn't just cosmetic — task types, pay structures, and onboarding flows have shifted:
- Image annotation remains the dominant task type. Self-driving and computer vision data is what Remotasks is really for now.
- Lidar and 3D annotation work continues — high precision, mid-tier pay.
- Text-based AI training (RLHF, coding eval) has largely migrated to Outlier proper. Remotasks doesn't really do this work anymore.
- Onboarding is now Outlier-style verification.
Pay ranges in 2026
- Image classification / bounding boxes: $5–$12/hr effective.
- Polygon segmentation: $10–$20/hr effective.
- Lidar / 3D annotation: $20–$40/hr effective.
- Specialty (medical imaging): $40–$80/hr effective.
The rates are lower than text-based AI training because the work is more mechanical. Image annotation specifically has been pressured by automation.
Who Remotasks still works for
- Visual-spatial workers. Some people genuinely prefer image work to text work; for them, Remotasks fits.
- Contractors in lower-cost markets. $10–$20/hr is competitive in many places.
- Specialty visual work. Medical imaging, lidar, autonomous-vehicle annotation pays meaningfully more.
- Backup hours. Steady availability complements text-based primary platforms.
Who should skip it
- US/EU-based generalists — text-based AI training pays much better.
- Coding contractors — Outlier coding tracks pay 3–5x more.
- Domain experts — your specialty earns multiples more on text platforms.
The bigger picture
Image-based AI training is a shrinking market in 2026. Frontier labs increasingly train multimodal models on existing image datasets plus synthetic data, with less reliance on human annotation. Computer vision-specific use cases (self-driving, medical imaging) still need human work, but generalist image annotation is being automated.
If you're in image annotation now and want to grow, the move is into specialty visual work (medical, lidar, 3D) or cross-train into text-based AI training.
Bottom line
Remotasks in 2026 is a niche platform for specific visual-spatial work. Generalist image annotation pays poorly compared to text-based AI training. Specialty visual work pays decently but applies to a narrow segment of contractors. Most readers should skip Remotasks and apply to Outlier or Mercor instead.