Use · Infercom

Is it good enough?

Ownership levelSubstantialnone·limited·partial·substantial·fullAnalytical input B · 74/100

This page is a projection of the one entry record, the Doesn't fail you factor that Use covers. The full verdict is set by all four factors together, floor-weighted so the weakest caps the whole.

Which domain expands which factor

AssessUse & modify + Transparency
ImplementData control + Doesn't fail you
UseDoesn't fail you
SupportTransparency

Models served

Infercom serves major open-weight models in native BF16 on SambaNova RDU dataflow hardware, across EU/US/JP regions. Serving in native BF16 (rather than a more aggressive quantisation) is a quality signal for fidelity-sensitive workloads. The exact model roster and version list were not fully confirmed in the public sources reviewed this session, so confirm the current catalogue and per-model availability with the provider before committing.

Inference features

The endpoints are OpenAI-compatible chat/completions, with native BF16 serving on dataflow hardware. Higher-level behaviours such as tool/function calling and structured (JSON) output depend on the specific served model's own capabilities and were not individually re-verified on Infercom's endpoints this session, so treat per-feature support as "inherited from the model, confirm per model."

How this scores

The ownership factor this domain covers, drawn from the one entry record.

Doesn't fail youDoes it stay up and stay secure?

Moderate

ISO 27001 and the CSA AI assessment give real security maturity, but the platform is newly launched (first datacenter late 2025) with no public uptime SLA or status history verified.

How this scores (AOI sub-dimensions)

Reliability2/5whether it stays up, with an SLA and status historySambaNova RDU dataflow claims strong performance/efficiency, but the platform is newly launched (first datacenter late 2025) and no public uptime SLA, status page or history was independently verified.

Security4/5the controls protecting your traffic and dataISO 27001-certified security management, a completed CSA AI security assessment and encryption in transit.

Compliance4/5which independent certifications and attestations it holdsISO 27001 certified - a third-party-audited standard - corroborated by the independent SambaNova sovereign-AI partnership announcement (ev3), which satisfies the independent-attestation bar for a claim the vendor cannot self-certify; plus a completed CSA AI security assessment (CSA STAR Level 1 self-assessment / AI CAIQ v1.1.0) and GDPR, German BDSG and EU AI Act alignment.

What this means for adoptionYou substantially own EU-sensitive workloads here: zero retention, prompts never trained on, metadata-only logs and a portable open-weight exit keep your data yours, and the posture is read from a binding Privacy Policy and an Art. 28(3) DPA rather than a marketing page. It stops short of full because sovereignty is conditional - Infercom's own Privacy Policy routes the default Global Model Catalogue's non-EU-hosted prompts to the US under SCCs - the platform is newly launched with no verified uptime SLA, and the sub-processor list is undisclosed. Pin EU-hosted models and sign the DPA to keep data in EU jurisdiction.

Sources

The same evidence records as the entry sheet. Read means the text was verified; unverified means it is known to exist but not yet read.

Vendor announcementunverified2026-07-25

Luxembourg-headquartered sovereign inference platform serving open-weight models on SambaNova RDU hardware with OpenAI-compatible APIs, EU sovereignty by default.

Vendor announcementread2026-07-25

Trust Centre (marketing surface): zero data retention, prompts never used for training, usage logs metadata only (timestamps/token counts/model, no content) with 90-day retention, ISO 27001 certified, CSA AI security assessment (CSA STAR Level 1 / AI CAIQ v1.1.0) completed, GDPR and EU AI Act alignment, encryption in transit.

Third-party analysisunverified2026-07-25

SambaNova partnership announcement - Infercom as a sovereign AI provider, Luxembourg HQ, EU jurisdiction, Munich deployment, ISO 27001, GDPR-safe/EU-compliant.

Third-party analysisunverified2026-07-25

SambaNova & Infercom Q&A - EU-hosted models on dedicated SambaNova RDU (SN40L) hardware at Equinix Munich; native BF16; three-tier memory; up to 10x faster / 5x more energy-efficient than GPU alternatives; first datacenter late 2025.

Third-party analysisunverified2026-07-25

Independent gateway partnership - open-weight models on SambaNova RDU dataflow hardware across EU, US and JP regions, OpenAI-compatible, no vendor lock-in.

Vendor announcementread2026-07-25

Zero Data Retention glossary - after inference completes the data is gone; prompts are never used for training.

Vendor announcementunverified2026-07-25

EU Sovereign AI page - EU-hosted models keep data in the EU; data never leaves the European Union.

Privacy Policyread2026-07-25

Privacy Policy (binding, clauses read this session): "Infercom does not store, log, or retain the content of your prompts or model outputs" - inference data is processed transiently, passed to the inference engine and the content discarded - and Infercom "does not use inference data for model training, fine-tuning, or service improvement" for both EU-Hosted Models and the Global Model Catalogue.

Data Processing Addendumunverified2026-07-25

Data Processing Agreement v1.3 - an Article 28(3) GDPR controller-processor agreement with Infercom SCS (Luxembourg-incorporated), effective 2026-03-04, imposing Articles 28-33 processor obligations.

Privacy Policyread2026-07-25

Privacy Policy (binding, clause read this session): for Global Model Catalogue models not hosted on EU infrastructure, "your API requests (including prompt content) are routed to SambaNova's global infrastructure outside the EEA, primarily in the United States"; Infercom uses EC Standard Contractual Clauses (Decision 2021/914) as the primary transfer mechanism, with EU-US Data Privacy Framework certification as an additional basis.