TypeSafe AI Launch

0:02

I'm Diego Almeida, founder of TypeSafe AI.

0:13

Our team trained the first models to be superhuman at instruction following.

0:24

But the trillion dollar question is, why not?

0:39

These flaws mean language models require humans in the loop, and almost no true automation can be done.

0:57

With a new architecture, a new sampler, and a new training algorithm.

1:18

Look at how the model generates sequentially.

1:53

Today, we're releasing Jev, the first public System One model.

2:06

A hundred times cheaper, with input tokens priced at forty two dollars per billion tokens.

2:32

Side effects may include fewer retries, fewer parsers, fewer broken schemas, and intelligence running inside every software loop.

2:55

As costs fall, usage may increase dramatically.