Neon outline of the map of the United Kingdom with a lightning bolt in the center, illuminated with purple and yellow lights inside a data center with server racks.

Frontier AI shouldn't mean foreign jurisdiction.

Most AI infrastructure forces a choice. You get frontier capability on US hyperscalers — and your data sits under US law. Or you get UK data residency on a sovereign cloud — and the models can't handle production workloads.

ADD is built differently. Dedicated inference capacity on UK-incorporated, UK-operated infrastructure. Frontier open-weight models including MiniMax M2.7. Your data never leaves the UK. Your stack stays under your control.

No CLOUD Act exposure. No surprise model deprecations. No vendor migration when pricing shifts.


The fastest infrastructure on earth for MiniMax M2.7

We run on SambaNova SN40L racks. Not Nvidia GPUs.

MiniMax M2.7 on SambaNova: 426.1 tokens/second.

Verified by Artificial Analysis across all production providers serving M2.7:

  • SambaNova: 426.1 t/s

  • Fireworks: 70.0 t/s

  • Together AI: 67.7 t/s

SambaNova also delivers the lowest time-to-first-token (8.12s vs 36+ seconds on next-fastest providers at 10K input token workloads).

Why SN40L:

  • Reconfigurable Dataflow architecture — fused kernels, no redundant memory calls

  • Three-tier memory (SRAM + HBM + DDR) — no memory bandwidth wall at inference scale

  • 10.2 PFLOPS per rack at full utilization

  • Air-cooled, standard 19-inch form factor

Purpose-built for inference. Not training hardware repurposed for serving.

A computer processor chip in a server room with purple lighting.

Simple, Scalable Pricing

Choose the plan that fits your data sovereignty and inference needs. All plans include UK-based infrastructure.

Free

Start exploring sovereign AI with free inference tokens. No credit card required.

Includes

  • 500,000 SovToks (Sovereign Tokens)
  • Access to MiniMax M2.7 + Non Sovereign models
  • No credit card required to start
  • Community support
  • UK data residency guarantee

Developer

Pay as you go for production inference that scales on-demand with UK sovereignty.

Includes

  • Pay-per-token pricing
  • Access to all production & preview models + Non Sovereign models
  • Standard rate limits
  • Standard support

Enterprise

Scale production traffic with dedicated infrastructure and advanced features.

Includes

  • Volume-based pricing & discounts
  • Access to all production & preview models
  • Dedicated infrastructure options
  • Priority support & SLA
  • Production rate limits

Dedicated Capacity

Your own reserved SambaNova racks, fully managed by us. Hosted in our UK datacenters with guaranteed capacity and priority support.

Includes

  • Guaranteed capacity
  • Custom model hosting
  • Performance SLAs
  • Priority support

Calculate your savings with MiniMax M2.7

Drag the sliders to see how much you'd save versus Claude Opus 4.7

MiniMax M2.7
Input£0.60 / M tokens
Output£3.00 / M tokens
Claude Opus 4.7
Input£3.90 / M tokens
Output£19.50 / M tokens
Input tokens per month 10M
1M250M500M750M1B
Output tokens per month 5M
1M250M500M750M1B
MiniMax M2.7 Cost
£21.00
per month
Claude Opus 4.7 Cost
£136.50
per month
You save
£115.50
85% lower cost