Sovereign model catalog

Qwen3.6-35B (NVFP4, GX10)

nvidia/Qwen3.6-35B-A3B-NVFP4
data residency · self-attested key custody · self-attested SEAL 0 · indicative
Front unknown

Qwen3.6-35B-A3B served on the self-operated GX10. Probe-verified tools + logprobs + sampling; served plain (reasoning off).

context
262,144
max output
8,192
pricing
1/3 EUR
per 1M (in/out)
endpoints
1

Get access

Every way to contract this model, ranked easiest-first. Prices in EUR per 1M tokens.

Capabilities

toolstext
languages
en · strongzh · nativees · strongde · supportedfr · supported

variants

  • floor
  • nitro
  • sovereign
  • extended

Endpoints

GX10 (self-operated, Madrid) · es/madrid-lan
Operator (self-operated GX10) (ES)
Tier 0 · self-attested SEAL 0 · indicative
Front unknown
runs in ES · self-attested
1/3 EUR · per 1M (in/out)throughput: 18.6 tok/slatency p90: 111msKey custody: customerstatus: live

no certifications claimed

SEAL objectives (SOV-1…8)

⚠ indicative — not an official EU SEAL assessment

ObjectiveWhat it asksLevelMethod
SOV-1
Strategic
Who ultimately owns and controls the operator?
SEAL-1self/contract
SOV-2
Legal & jurisdictional
Which law governs the operating entity — and can a non-EU government compel it?
SEAL-1self/contract
SOV-3
Data & AI
Where do the data, keys and inference actually live?
SEAL-1self/contract
SOV-4
Operational
Who runs and observes the service day to day?
SEAL-2probe
SOV-5
Supply chain
Where does the hardware and silicon come from?
SEAL-0 · contradictedhardware-disclosure
SOV-6
Technology
How independent is the underlying stack?
unknown — not assessed—
SOV-7
Security & compliance
Which certifications actually hold?
unknown — not assessed—
SOV-8
Environmental
Is the energy and sustainability posture accounted for?
unknown — not assessed—

Overall = lowest assessed objective; ≥3 unknown → no grade. · SEAL 0 · 3/8 unknown

Use it (OpenAI SDK, sovereignty-pinned)

Copy-runnable: the real OpenAI SDK pointed at NeuralRing with a DE jurisdiction pin.

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://neuralring.eu/api/v1",
  apiKey: process.env.NEURALRING_API_KEY,
});

const res = await client.chat.completions.create({
  model: "nvidia/Qwen3.6-35B-A3B-NVFP4",
  messages: [{ role: "user", content: "Hello from a sovereign endpoint" }],
  // NeuralRing extension — the sovereignty pin is applied FIRST and never loosened:
  // @ts-expect-error sovereignty is a NeuralRing routing extension
  sovereignty: { jurisdiction: ["DE"], min_assurance: "self_attested" },
});
console.log(res.choices[0].message.content);