Proximal

Long-horizon coding RL environments built from real codebases

Proximal builds long-horizon software engineering environments from real codebases for RL training of coding agents at frontier labs. It publishes the FrontierSWE benchmark.

FrontierSWE benchmark

Websiteproximal.ai
DomainsCoding, Long Horizon
LocationSan Francisco, Bangalore
Team size26-50
FoundersJustus Mattern, Navid Pour, Calvin Chen

Similar startups

Also listed under Coding, Long Horizon.

NameDomainsLocationTeamFundingRaising
Rise Data Labs
US-based expert human data and custom RL environments for enterprise AI
Computer Use, Coding, Finance, Cybersecurity, Legal, Data Labeling, Enterprise, Multi-Domain, RLHF, Custom EnvironmentsUnited States--Yes
AfterQuery
Expert human data and RL environments across code, finance, and computer use
Multi-Domain, Coding, FinanceSan Francisco, New York, Seattle51-100$30.5M total (reported)-
AIChamp
Custom enterprise-workflow RL environments with expert grading
Enterprise, Long Horizon, Custom EnvironmentsSan Francisco1-10--
Akhara
Enterprise and code RL environments
Enterprise, CodingSan Francisco1-10--
Andon Labs
Long-horizon autonomy benchmarks like Vending-Bench
Long Horizon, AlignmentSan Francisco1-10Seed (Y Combinator)-
Andromede
Programmatic generation of long-horizon RL environments
Long HorizonLausanne1-10--