primis beta update: we’ve been rebuilding primis compute around one clear lesson from builders in our closed beta: the problem is not discovering more compute. the problem is knowing what compute will cost, trusting that rate, and being able to build around it. when compute becomes a core production input, pricing errors, latency, and volatility are not small issues. they hit margins, reliability, and product velocity. this is why we’re rebuilding the rate engine from the ground up. from the builder side, open beta is being shaped around one simple flow: query rate -> reserve rate -> attach workload -> reconcile usage no provider noise. no new workflow to learn. no extra complexity for builders. primis handles the rest behind the scenes. predictable compute pricing inside the stack builders already use. primis mode.