Enterprise Offering

Frontier AI coding, hosted in your sovereign region.

A dedicated GPU node inside your jurisdiction, running frontier open-weight coding models. Unlimited agentic coding for your developers — and token data that never leaves the hardware.

~1.5B
tokens a single developer on agentic tools burns per month
CHF 5K
what that costs per developer, every month, at frontier API rates
62%
of EU organisations seeking sovereign solutions — 76% in banking
2.2 pts
gap between the best open-weight model and Claude Opus 4.7 on SWE-bench Pro

Your developers need frontier AI. Your compliance team needs control.

Two problems, one answer: a dedicated node, in your region, running open weights.

The token tax

~CHF 5K / developer / month

A single developer on agentic tools burns roughly 1.5 billion tokens a month. At frontier API rates that is around CHF 5,000 per developer, every month. Per-token pricing does not survive at team scale.

  • closeMetered on every request, however routine
  • closeHard rate limits exactly when the team pushes hardest
  • closeA monthly bill nobody can forecast

The sovereignty risk

62% of EU orgs, 76% in banking

European organisations are seeking sovereign solutions. The US CLOUD Act reaches US-owned clouds even inside an EU datacenter. Sending source code to a US API is an exposure you cannot audit.

  • closeSource code leaves your network on every request
  • closeUS CLOUD Act reaches US-owned clouds, even in-region
  • closeNo audit trail for where prompts actually go

Both point to the same answer: a dedicated node, in your region, running open weights.

Your own dedicated AI coding server, in your region.

A reserved GPU node provisioned to your jurisdiction, serving frontier open-weight models over a standard OpenAI-compatible endpoint — with a telemetry-only control plane.

public

Data Residency by Region

Every node runs on dedicated in-region GPU infrastructure. You pick the region; the node — and all inference on it — stays there. Switzerland is the first region available, with more to follow.

shield

Prompts & Completions Never Leave the Node

Requests go straight to your dedicated hardware over TLS. The node validates keys locally and serves inference locally. No central proxy sits in the request path.

lock

Telemetry-Only Uplink

The only data that reaches syndicAI is health checks and aggregate usage counts — never prompts, completions, code context, or per-request content. Our control plane cannot read your traffic.

dns

Always-On, Single-Tenant

A permanent node reserved for one organisation. No shared tenancy, no spot reclamation, no idle-stop surprises. It's your box, and only yours.

Point your existing tools. Nothing leaves your node.

One base URL and one key. The request path ends on hardware you pay for — not on a foreign cloud.

terminal

Your developers

IDEs and agents

The tools your team already uses — Cursor, Continue, aider, any SDK — configured with one base URL and key. No rewrite.

arrow_downward
api

OpenAI-compatible

One endpoint

A drop-in /v1/chat/completions API. TLS terminates on the node itself, and keys are validated locally against synced hashes.

arrow_downward
dns

Your dedicated node

GPUs in your region

Reserved GPU hardware running the model. Prompts, completions, and logs stay here — pinned to the jurisdiction you chose.

check Token data never leaves the node check Drop-in OpenAI-compatible API check Telemetry only, never payloads

Frontier open-weight models, fully managed.

Every model ships open-weight and open-source licensed. We manage the serving stack, updates, and tuning — you own the hardware it runs on.

Agentic All-Rounder

MiniMax M3

MiniMax · MoE · 1M context

Frontier coding plus native multimodal and computer use — built for long-horizon autonomous agents.

Open-Weight Leader

GLM-5.2

Z.AI · ~750B MoE · MIT license

The strongest open-weight coder available. Deep reasoning and tool use across full repositories.

Code Specialist

Kimi K2.7 Code

Moonshot AI · ~1T total · 32B active

Maximum capability for the hardest, longest-horizon agentic and coding tasks, with ~30% fewer reasoning tokens than its predecessor.

A flat monthly price, not a token meter.

You are buying capacity, not a meter. The price is the price, however hard your team pushes it.

Per-token API

Metered, throttled, off-site

The default today: every request billed per token, rate-limited at peak, and served from someone else’s cloud.

  • closeMetered per token, on every request
  • closeHard rate limits at peak load
  • closeSource code leaves your network
  • closeA monthly bill nobody can forecast

Your syndicAI node

Flat, unlimited, on your hardware

A dedicated node with a fixed monthly subscription. Push it as hard as you want — the bill does not move.

  • checkOne flat monthly price
  • checkNo rate limits, unlimited tokens
  • checkData stays on your hardware
  • checkA budget you can set a year ahead

Dedicated In-Region GPU, Fully Managed

The novel infrastructure — direct routing, on-node auth, subscription billing — handled for you.

lan

Direct Routing

Each node terminates TLS itself and serves on its own endpoint. Requests reach your hardware directly — no shared control-plane proxy in front of your traffic.

key

On-Node Key Auth

API keys are validated on the node against synced hashes with a revocation list. Keys can be rotated and revoked without exposing plaintext to anyone.

model_training

Frontier-Class Open Weights

Run MiniMax M3, GLM-5.2, Kimi K2.7 Code and more on 96 GB-per-GPU hardware, scaled from 1 to 4 GPUs via vLLM tensor parallelism.

api

OpenAI-Compatible API

A standard /v1/chat/completions endpoint. Point Cursor, Continue, aider, or any SDK at the node URL and go.

receipt_long

Predictable Subscription Billing

A flat monthly price per node, tiered by GPU count and size. No per-token metering, no spot-price swings, no surprise overage.

workspaces

Organisation Workspace

Manage nodes, members, API keys, and your subscription from a dedicated enterprise workspace — separate from the spot squad experience.

Deployed instantly. Live in your region within the hour.

Onboarding is four self-serve steps — no procurement maze. Deployment starts immediately; provisioning the dedicated node takes up to an hour. Switzerland first; further regions on request.

1

Scope

Pick a model, region, and team size. We size the node to your workload.

2

Provision

Deployment starts instantly. The reserved node comes up inside your jurisdiction within the hour, to a fixed spec.

3

Connect

Developers point existing tools at the OpenAI-compatible endpoint.

4

Run

Unlimited agentic coding. We manage model updates and uptime.

Enterprise FAQ

Let's provision your node.

Spin up a dedicated, single-tenant node where your prompts and completions never leave the machine. Switzerland available today — further regions on request.