Announcing African Region Nodes

Latency is not a small problem. When your AI gateway is sitting in a US-East data centre and your users are in Lagos or Nairobi, you're adding 200–300ms of round-trip overhead before the model has even seen the request. For a streaming chat response, that first-token delay is what users feel. It's the difference between a product that feels alive and one that feels like it's thinking too hard.

Today we're announcing the rollout of AmgapTech region nodes across Africa. Starting with nodes in West Africa (Lagos) and East Africa (Nairobi), with Southern Africa (Johannesburg) following in Q2.

Routing through a regional node brings average first-token latency down from ~280ms to ~60ms for requests originating in those regions. We handle the intelligent routing — your API calls don't change. Point at our gateway, set your preferred region in the header, and we do the rest.

Honestly? Infrastructure costs. Running redundant nodes in African data centres is more expensive per request than US or EU regions, and the colocation options are fewer. We held off until we could guarantee the same uptime SLA (99.9%) rather than ship something that would fall over under load. We think that was the right call.

Regional pricing will reflect the actual infrastructure cost — slightly higher than our base tier, but significantly cheaper than what you'd pay routing everything internationally. Details on the pricing page.