X · @togethercompute
· X / Twitter
Running open-weight models in production gets hard fast. On August 6, we’re walking through major updates to our inference platform that make deploym…
Running open-weight models in production gets hard fast.On August 6, we’re walking through major updates to our inference platform that make deployment and operations easier:→ Safe rollouts: canary, blue-green, automatic rollback→ A/B and shadow testing on live traffic→ SLO-driven autoscaling→ Deploy any model with ~4x