TypeSafe

[San Francisco, California, USA]typesafe.ai

TypeSafe describes itself as an AI lab building machine-native intelligence infrastructure for automation, designed to make decisions within software. It is led by CEO Diogo Almeida, who co-invented RLHF and InstructGPT, with COO Sasha Sheng (previously Meta FAIR) and CTO Erik Gafni. Its models are trained with RLCD, which the lab defines as reinforcement learning for calibrated decisions, so they return typed answers and probabilities rather than generated text.

sources:typesafe.ai typesafe.ai/…/team docs.typesafe.ai/…/machine-learning-primer docs.typesafe.ai/…/system-one

· verified 2026-09-20

models

family

■ decision 1

1 models · 1 family

Jev

newest: Jev 1.13

[product][api]

Family

Jev

[product][api]

1 in catalog

Jev 1.13

First System One model: choice, score and noul answers with calibrated probabilities, 64k context, text-only state

[decision]64k ctxnewest

sources:docs.typesafe.ai