Transparent Pricing. Every Service. Every Scale.
Three engagement models designed for AI teams at every stage. Every project starts with a free pilot so you see the quality before spending anything.
Three Ways to Work With Us
Every engagement starts free. Scale to the model that fits your pipeline.
- ✓Up to 500 labeled items on your actual data
- ✓Your rubric, your format, your output schema
- ✓48hr turnaround standard
- ✓4-layer QA pipeline applied in full
- ✓QA accuracy report on delivery
- ✓Applies to all service types
- ✓500 to 50,000+ items per batch
- ✓Dedicated project lead assigned
- ✓All six LLM eval service types supported
- ✓Full 3D LiDAR and 2D CV pipeline available
- ✓Disagreement reporting with rationale per batch
- ✓Weekly QA reports and calibration checks
- ✓NET-30/60/90 invoicing on request
- ✓Volume discounts at 10K, 50K, 100K items
- ✓5 to 30 raters, your pipeline only
- ✓Continuous throughput, no batch delays
- ✓Priority 24 to 48hr turnaround SLA
- ✓All service types including multilingual
- ✓Real-time QA visibility on request
- ✓MSA and custom NDA available
- ✓Quarterly business reviews
All projects start with a free pilot. Pricing confirmed after reviewing your data and rubric. No rate card surprises mid-pipeline.
Not sure which model fits your pipeline?
Book a 15-minute call and we will walk through your requirements, volume estimates, and timeline. We will recommend the most cost-effective approach for your specific use case.
Pricing by Service Type
Per-unit pricing varies by annotation complexity, scene density, and object class count. Rates below are typical ranges. Volume discounts apply automatically at 10K, 50K, and 100K items.
| Service | What It Covers | Per-Unit Range | Throughput |
|---|---|---|---|
| LLM Evaluation & Human Judgment (Primary Lane) | |||
| RLHF Preference Pairs | Pairwise response ranking for LLM fine-tuning | $0.50 – $8.00 per pair | 1,000 – 3,000 pairs/week |
| Agent Benchmark Evaluation | Multi-step task completion and tool-use scoring | $5.00 – $25.00 per task | 200 – 800 tasks/week |
| Audio CSAT Evaluation | Naturalness, accuracy, and satisfaction scoring | $6.00 – $25.00 per session | 300 – 1,000 sessions/week |
| LLM Output Labeling | Quality, safety, factual accuracy, instruction-following | $0.50 – $5.00 per item | 2,000 – 8,000 items/week |
| Multilingual Evaluation | FR, ZH, DE rater capacity on any eval type | +30 – 60% premium on base rate | By language pool availability |
| Red-Team Adversarial Testing | Structured per-session with failure taxonomy reporting | Custom per session | Scoped per engagement |
| 3D LiDAR & Sensor Fusion (Computer Vision Lane) | |||
| 3D LiDAR Cuboids | Object detection with position and orientation | $0.20 – $1.50 per object | 100 – 500 frames/week |
| Point Cloud Segmentation | Per-point semantic and instance labeling | $5.00 – $25.00 per frame | 50 – 200 frames/week |
| Multi-Sensor Fusion | LiDAR + camera + radar alignment | $8.00 – $35.00 per frame | 50 – 150 frames/week |
| Temporal Tracking (3D) | Object ID consistency across sequences | $15.00 – $50.00 per sequence | 20 – 100 sequences/week |
| AV Edge Case Annotation | Weather, occlusion, rare objects | $0.50 – $3.00 per item | 500 – 2,000 items/week |
| Robotics / 6DoF Pose | Pose estimation, grasp annotation | $2.00 – $10.00 per object | 200 – 800 objects/week |
| 2D Image & Video Annotation | |||
| Bounding Boxes | Standard objects, clear visibility | $0.015 – $0.035 per object | 2,000 – 5,000 images/week |
| Polygon Segmentation | Complex shapes, precise boundaries | $0.03 – $0.10 per object | 1,000 – 3,000 objects/week |
| Video Object Tracking | Multi-object, temporal consistency | $2.00 – $8.00 per video | 100 – 400 videos/week |
| Keypoint Annotation | Pose estimation, COCO 17-pt or custom | $0.012 – $0.020 per keypoint | 3,000 – 8,000 frames/week |
3D vs 2D pricing: 3D LiDAR annotation runs 5 to 25x more per frame than 2D bounding boxes. This reflects spatial complexity, specialized training, and annotation speed (3D is 3 to 10x slower than 2D). Our Nairobi-based team delivers a 40 to 60% cost advantage over Western providers at equivalent quality.
Everything Comes Standard
No hidden fees, no extras billed separately. Every project includes the full quality and delivery infrastructure below.
Quality Assurance
- ✓4-layer QA: self-review, peer check, lead audit, client-spec validation
- ✓98.5% accuracy standard across all service types
- ✓Inter-annotator agreement measured per batch
- ✓Disagreement flagged with rationale, not just a score
- ✓3D-specific: IoU validation, position and orientation accuracy checks
Project Management
- ✓Dedicated project lead on every engagement
- ✓Custom rubric training before production starts
- ✓Weekly progress reports and calibration checks
- ✓Guideline versioning with 24hr adjustment turnaround
- ✓First revision round included at no extra cost
Data Security
- ✓NDA standard on every project before data transfer
- ✓Self-hosted infrastructure, no third-party SaaS exposure
- ✓ISO 27001 in-flight, target Q4 2026
- ✓GDPR posture in progress, DPO in-flight
- ✓Role-based access, audit trails, daily encrypted backups
Delivery and Formats
- ✓LLM eval: JSON, JSONL, CSV, or custom schema
- ✓3D: KITTI, nuScenes, Waymo Open, Datumaro, custom
- ✓2D: COCO JSON, YOLO, Pascal VOC, custom schemas
- ✓Versioned datasets with full revision history
- ✓Post-delivery integration support on request
Common Questions
How does the free pilot work?
We annotate up to 500 items from your actual dataset at zero cost. You receive the full output with a QA report showing accuracy metrics. No payment information required. Applies to LLM evaluation, 3D LiDAR, and 2D annotation equally.
What determines per-unit pricing for LLM evaluation?
Task complexity, rubric depth, required turnaround, and language. A simple preference pair on a clear prompt costs less than a multi-turn agent evaluation requiring external state verification. We scope accurately after reviewing your rubric and sample data.
What determines per-unit pricing for 3D LiDAR?
Objects per frame, scene density, annotation type (cuboids vs. segmentation vs. tracking), number of classes, attribute depth, and whether multi-sensor fusion is involved. We review sample frames during consultation before quoting.
Why is 3D annotation more expensive than 2D?
3D annotation requires spatial reasoning across multiple viewports, specialized training (120+ hours per rater), and runs 3 to 10x slower than 2D. Our Nairobi team delivers a 40 to 60% cost advantage over Western providers at equivalent quality.
Is pricing locked once we start?
Yes. Project-based pricing is locked for a 180-day project window. No rate surprises mid-pipeline, no renegotiations after volume increases. Dedicated retainer pricing is fixed monthly for the contract term.
What are your payment terms?
Project-based: 50% upfront, 50% on delivery for projects under $10K. NET-30 for projects over $10K or established clients. Dedicated retainer: monthly invoicing with NET-15 or NET-30 terms. Wire transfer, Payoneer, and standard invoicing accepted.
Do you work with early-stage companies?
Yes. The free pilot exists precisely for this. You verify quality before committing budget. Flexible project sizing and payment terms available for teams with limited runway. No minimum project size.
Can you handle confidential or sensitive data?
Yes. NDA is standard before any data is transferred. Self-hosted infrastructure means your data never touches third-party SaaS. ISO 27001 in-flight, GDPR posture in progress. MSA and custom NDA available on the dedicated retainer.
Start With a Free Pilot
200 to 500 items. No cost, no commitment. You see the quality on your actual data before any contract is signed.