Tokenless

Active · We find flaws in RL environments and evals and help you fix them.

YC S26

Website YC profile

Why it ranks

How the rank works

7signal of 100 · rising 7

went through YCProgram · this site+4
team of 3Team size · YC's directory+3.1

A sum of public facts, not a judgement and not investment advice.

Competitors

EpsilabShared: research lab, reinforcement learning, machine learningYC F26San FranciscoStartup →AI

Vibrant LabsShared: reinforcement learning, eval, environmentYC W24San FranciscoStartup →AIAcquired

Aquarium LearningShared: model performance, machine learning, fixYC S20San FranciscoStartup →AI

Synthetic SciencesShared: research lab, reinforcement learning, evalYC W26San FranciscoStartup →AI

idlerShared: research lab, reinforcement learning, evalYC S25San FranciscoStartup →

Companies closest to Tokenless
Company	Category	Status	Program
EpsilabShared: research lab, reinforcement learning, machine learningYC F26ActiveAI, Data	AI, Data	Active	YCF26
Vibrant LabsShared: reinforcement learning, eval, environmentYC W24ActiveAI, Devtools	AI, Devtools	Active	YCW24
Aquarium LearningShared: model performance, machine learning, fixYC S20AcquiredAI, Devtools	AI, Devtools	Acquired	YCS20
Synthetic SciencesShared: research lab, reinforcement learning, evalYC W26ActiveAI, Devtools	AI, Devtools	Active	YCW26
idlerShared: research lab, reinforcement learning, evalYC S25ActiveAI, Devtools	AI, Devtools	Active	YCS25

Adjacent

AIInactive

NuntiusShared: reinforcement learning, model, alignmentYC S25San FranciscoStartup →AI

AfterQueryShared: research lab, model, appliedYC W25San FranciscoStartup →AI

AnchorheadShared: reinforcement learning, eval, modelYC W26San FranciscoStartup →AI

RightNowShared: research lab, machine learning, modelYC F26San FranciscoStartup →AI

TraverseShared: reinforcement learning, machine learning, modelYC W26San FranciscoStartup →

Companies next to Tokenless
Company	Category	Status	Program
NuntiusShared: reinforcement learning, model, alignmentYC S25InactiveAI, Devtools	AI, Devtools	Inactive	YCS25
AfterQueryShared: research lab, model, appliedYC W25ActiveAI, Data, Devtools	AI, Data, Devtools	Active	YCW25
AnchorheadShared: reinforcement learning, eval, modelYC W26ActiveAI, Devtools	AI, Devtools	Active	YCW26
RightNowShared: research lab, machine learning, modelYC F26ActiveAI, Devtools	AI, Devtools	Active	YCF26
TraverseShared: reinforcement learning, machine learning, modelYC W26ActiveAI	AI	Active	YCW26

Also from S26 in AI

All of S26

Autoresearch, Inc.Autoresearch as a service: we solve your hardest measurable problemsYC S26San FranciscoStartup →AI

Agnost AIWe find where AI agents fail, then train better models to run them.YC S26San FranciscoStartup →AI

CoArenaThe biggest crowdsourced benchmark for Computer-UseYC S26San FranciscoStartup →AIInactive

MentlioEngineering Intelligence and Token Optimization for the AI-Coding EraYC S26San FranciscoStartup →AI

RiftenEarned intelligence. An independent AI research lab.YC S26San FranciscoStartup →AI

ArchalAPI sandboxes, built for AI agentsYC S26San FranciscoStartup →

Company	Category	Status
Autoresearch, Inc.Autoresearch as a service: we solve your hardest measurable problemsteam of 3ActiveAI, Devtools	AI, Devtools	Active
Agnost AIWe find where AI agents fail, then train better models to run them.2 programsteam of 2ActiveAI, Data, Devtools	AI, Data, Devtools	Active
CoArenaThe biggest crowdsourced benchmark for Computer-Useteam of 2ActiveAI, Data, Devtools	AI, Data, Devtools	Active
MentlioEngineering Intelligence and Token Optimization for the AI-Coding EraInactiveAI, Devtools	AI, Devtools	Inactive
RiftenEarned intelligence. An independent AI research lab.team of 5ActiveAI, Devtools	AI, Devtools	Active
ArchalAPI sandboxes, built for AI agentsteam of 2ActiveAI, Devtools	AI, Devtools	Active

Founders

Rohit AgarwalFounder
Andrew LiuFounder
Kevin WangFounder

About Tokenless

Tokenless is an applied AI research lab. Our goal is to end reward hacking. We do research and build tools to detect issues with environments and evals, understand model behavior in those environments, and prevent those issues from happening. Correcting them makes training more effective, makes model performance easier to interpret, and is crucial to solving the AI alignment problem. If you’re training models with RL, building environments, or running agent evals, and want more confidence in your environments, we're here to help.

YC batch: S26 · Summer 2026
Status, per YC: Active
Category: AI, Developer tools
Team size: 3
Location: San Francisco, CA, USA
Founded: 2026
Checked: 6h ago

YC tags: Developer Tools, Machine Learning, Reinforcement Learning, B2B, AI