Tokenless EnvCheck

Developer tool

Tooling and expertise for finding reward hacks and grading bugs in reinforcement-learning environments and agent evals, including the EnvCheck audits and the Eval³ leaderboard of how well models audit benchmarks, for teams training models with RL, building environments or running agent evals; the home page asks interested teams to get in touch and publishes no prices.

Find alternatives to Tokenless EnvCheck

TypeDeveloper tool

RoleStandalone tool

Bought forTest and evaluate an AI agent, Train or fine-tune a model

AvailabilitySold

DeploymentSaaS

Intended forDeveloper, Researcher

SecurityNot recorded

StatusActive

Sold withinPackaging not stated — the vendor’s page was checked and named no parent or plan

PricingEnterprise quote

Pricing, Deployment, Sold within — usetokenless.com · 20 Oct 2026

What it does

Evaluation and observability
Model training

Features 4

Environment and eval auditingProvides tooling and expertise to understand and prevent failures in tasks and graders.
EnvCheckPublished a post introducing EnvCheck, which found 110 reward hacks and grading bugs in three public benchmarks.
Eval³ leaderboardPublishes a leaderboard of how good models are at benchmark auditing.
For RL training teamsAimed at teams training models with RL, building environments or running agent evals.

usetokenless.com