FrontierRL
0,001

Tokens / second

Stop rationing

Frontier Inference

Decoding throughput, tokens per second.

4X faster than the frontier lead

Star outputs measured on workstation GPUs, not datacenter-grade.

At Budget-Level Prices

Calculated per 1M output tokens, at each model’s published price.

How far does $100 go?

~75× more tokens per dollar

Watch Star run.

Reacting in real timeSmash 64 Halo, no trainingHalo Sourcing LinkedIn candidatesThe Web Playing three weeks, unattendedDiablo II Rebuilding the MIT RacecarCAD Autodesk Beat Pokémon for $6Pokémon Designing PresentationsPowerPoint 6 agents working togetherStarCraft II

Why Star is different

Not another LLM.
A new architecture, built for agents
from day one.

Actually parse live video, audio, text, data and act in real time.

Built by engineers from

MITAmazon
Immunai

Programs

TNT MIT TNT Research Together.ai Together.ai Accelerator

Stop rationing.

See the metrics