UK sovereign LLM inference

· ai coding · Source ↗

TLDR

  • RelaxAI (a Civo spin-out) offers OpenAI-compatible inference API via Python, TypeScript, Go SDKs and REST, targeting UK sovereign workloads.

Key Takeaways

  • Drop-in OpenAI compatibility: swap to pip install relaxai or npm install relaxai with no client code changes.
  • REST endpoints expose NLP models with documented request parameters and response schemas, designed for standard HTTP tooling.
  • Go SDK available via go get github.com/relax-ai/go-sdk@latest, covering three official language targets at launch.
  • Positioned for workloads requiring in-country data processing, not as a general-purpose hyperscaler.

Hacker News Comment Review

  • The landing page provided almost no company, datacenter, or model provenance information at launch; commenters had to extract basics from founder replies in the thread.
  • Founder confirmed: wholly UK-incorporated, spun out from Civo (managed Kubernetes provider, est. 2018, founded by Mark Boost) in 2026, UK-resident founders, not a foreign subsidiary.
  • Sovereign AI framing is gaining traction as a real procurement signal, not just marketing, driven by US soft-power retreat and allied nations prioritising in-country capability.

Notable Comments

  • @mrdw: pypi.org/project/relax is an abandoned package; owner email domain expired, so the name may be claimable – but follow PEP 541, not domain takeover.
  • @kybernetikos: Suggests UK sovereign AI strategy should aggressively recruit from the recent Qwen core team as a concrete capability shortcut.

Original | Discuss on HN