The RL data engine for AI teams

From environments to custom evaluations, we partner with over 90% of leading AI labs in the US and the innovators defining the next frontier of AI.

For frontier AI

The data, environments, and evaluation infrastructure the world's frontier AI labs build on.

Simulation environments

Tuned RL environments and preference signals for RL post-training and evals — covering reasoning, tool use, and computer use across the highest economic value domains: autonomous AI research, scientific knowledge work, agent coding, and cybersecurity.

Robotics

Full-stack data products for robotics foundation models. Video, trajectories, and rich multimodal annotations — collected with purpose-built hardware and AI-powered diversity engines, from pre-training through post-training to evals.

Alignerr

Real-world grounding signal from the smartest people on earth — the kind frontier models can't generate themselves. Alignerr brings together 2.6M+ knowledge experts, channeling human endeavor into the reward signals frontier AI needs to be useful.

For enterprises

Build it. Evaluate it. Deploy it.

Agent Studio is the platform for building, evaluating, and continuously improving specialist AI agents on real enterprise workflows at a fraction of frontier AI token costs.

Agent Studio

The platform for enterprise agents that work in production.

Step 01

Data integration

Connect enterprise APIs, SaaS tools, databases, and internal systems into production-grade simulation environments

Step 02

Scenario generation & grading

Design agent tasks with expert-built scenarios, synthetic edge cases, and structured rubrics for outcome and process evaluation

Step 03

RL training, serve & improve

Close the loop with RL training on evaluation reward signals, deploy production-ready agents, and continuously capture edge cases

Hundreds of AI teams build with Labelbox

How Meta built GIM with Labelbox data to evaluate frontier AI reasoning

Problem

Meta needed a benchmark that remained discriminative as existing LLM evaluations saturated. The team wanted tasks grounded in practical reasoning rather than obscure knowledge or synthetic puzzles, with enough rubric detail to capture partial credit and enough quality control to support a public-private contamination diagnostic.

Solution

Labelbox produced the data foundation for GIM: 820 expert-authored problems across seven cognitive categories, including 229 multimodal items and 528 rubric-graded prompts. The work included original prompt creation, structured scoring criteria, review, annotation, and quality assurance, enabling Meta to calibrate a 2PL IRT model over more than 200,000 prompt-response pairs.

Result

Meta released GIM-615, calibrated item parameters, and an evaluation framework that benchmarked 22 models across 47 reporting configurations. The paper found GIM remains far from saturated, with roughly 20% of items above frontier ability, giving researchers a durable way to compare model capability, thinking budgets, and future systems.

Technology and software

Labelbox Research

Latest work from Labelbox Research

Labelbox's world-class applied research team pioneers frontier AI data generation and evaluation methods. Through scientific precision and co-innovation, we help customers achieve real-time AGI breakthroughs.

EchoChain

An audio benchmark for reasoning under pressure in full-duplex dialogue

Implicit intelligence

Evaluating agents on what users don’t say

Intent laundering

The AI safety illusion: why current safety datasets fool us on model safety