Glossary
Staged release
Publishing a model's weights only after a sequence of gated steps, typically API access, external safety testing and red-teaming, rather than on launch day.
Because open weights cannot be withdrawn once downloaded, some labs now treat publication as the last step rather than the first. Thinking Machines' framework, used for Inkling 2 Small, makes each stage conditional on evidence from the previous one. Critics note it delays independent evaluation; supporters argue it is the only way to combine open weights with a credible safety process.
Related terms
AlignmentClosed modelOpen-weight modelWeights
Get the brief in your inbox
One email every morning: the stories that mattered in AI, each with a one-line "why it matters" for your role. Free, no ads, unsubscribe any time.
Published daily by AI Briefing. We never share your address.