Markets
Seams Collapses Gateway, Metering and Spend Caps Onto One Request Path and Kills a Stream at 98% of Budget
Seams, posted to Show HN on September 9, sits under an AI product as a single layer doing model access control under branded names, real-time cost metering as tokens stream, and mid-stream enforcement that stops a response at 98% of the customer's allocated budget. It explicitly targets the stack most teams assemble from OpenRouter or LiteLLM for routing plus Helicone or Portkey for observability plus a separate billing system, and ships a branded end-customer portal for usage, spend, keys and activity. Killing a response mid-stream rather than reconciling overage after the fact is the architectural claim worth watching; pricing was not disclosed on the page.
Source
↳ Follow the thread