Compare vendors / Abundant
Abundant
Builds coding environments and datasets for RL
Coding and SWE
Abundant creates environments and datasets for RL and also offers data licensing and custom evals. Its SWE-Marathon benchmark (NeurIPS 2026 poster) has 20 ultra-long-horizon tasks at about 27M tokens per attempt, with reward hacking observed in 13.8% of rollouts. It is a Y Combinator company backed by angels from OpenAI, Anthropic and Google DeepMind.
Best for: Ultra-long-horizon coding tasks with published reward-hacking analysis
1 Legal nameAbundant
2 HeadquartersSan Francisco, CA
3 Founded2024
4 Team1–10
5 DeliveryBespoke builds, Catalog licensing
6 ExclusivityNot disclosed
7 Customers and partnersSays it powers "every frontier AI lab and multiple neolabs"; names none.
8 Published pricesNot disclosed
Worth knowing before you sign
- Broad lab claim with no named customers
- No disclosed funding or named investors
- Founders not named on its own website
Sources
- Abundant homepage
- Y Combinator profile
- Built In profile
- SWE-Marathon paper (arXiv)
- pavlovslist directory
Work at Abundant? Send corrections with a source to editors@toprlenvironments.com. We fix verified errors within five business days.
Compare with
- Prime Intellect Runs the Environments Hub, open-source RL tooling and hosted RL training
- HUD Platform and marketplace for building, buying and running RL environments
- General Reasoning Hosts RL environments as API endpoints through OpenReward
- Veris AI Simulated copies of a company’s systems for testing and training enterprise agents
- Vmax Generates new environments automatically from your data and evals