Nebius Customer AtlasNot affiliated · Research by Not affiliated · By @anthonybs
Not affiliated with Nebius · Independent research by @anthonybsSee changes
AllOrbiq

Orbiq

Vendor assurance and compliance software

How they use Nebius Token Factory

Orbiq describes Nebius Token Factory as its primary inference provider after a Mistral outage interrupted a customer demo. It selected the service for EU infrastructure, zero data retention and open-model pricing. Its tasks include language detection, question parsing, compliance answer generation, deep search and questionnaire evaluation.

The implementation assigns Qwen models to different tasks, including Qwen3-14B for language detection and Qwen3-235B variants for answer generation and reasoning. Each AI call passes through a provider chain that checks outage status, retries failed calls and falls back to Mistral. Orbiq reports a 75% reduction in reasoning costs after rebuilding the architecture and changing its model assignments.

Sources

2 sources

How a Demo Outage Forced Us to Rebuild Our Entire AI Inference Architecture

4 Mar 2026 · Orbiq · Web page

Orbiq describes primary inference on Nebius Token Factory, task-specific Qwen models, Mistral fallback and a reported 75% reduction in reasoning costs.

Open original source
How a Demo Outage Forced Us to Rebuild Our Entire AI Inference Architecture4 Mar 2026 · Orbiq · Web pageOrbiq describes primary inference on Nebius Token Factory, task-specific Qwen models, Mistral fallback and a reported 75% reduction in reasoning costs.

Emre Salmanoglu on Orbiq's inference architecture

4 Mar 2026 · LinkedIn · Web page

Emre Salmanoglu describes Orbiq's rebuilt inference architecture with Nebius as primary, multiple-provider fallback and reasoning costs reduced by 75%. The post links Orbiq's technical article.

Open original source
Emre Salmanoglu on Orbiq's inference architecture4 Mar 2026 · LinkedIn · Web pageEmre Salmanoglu describes Orbiq's rebuilt inference architecture with Nebius as primary, multiple-provider fallback and reasoning costs reduced by 75%. The post links Orbiq's technical article.

Company context

orbiqhq.com

Orbiq develops trust center and vendor assurance software for B2B companies. Security, compliance, procurement and legal teams use its platform to share security documentation with buyers and auditors, collect vendor evidence and manage third-party assessments. Its Trust Center Platform provides branded portals with public, restricted and NDA-gated documents. Other features include AI-assisted security questionnaires, vendor evaluations, continuous monitoring and AI search across published compliance content. The platform connects existing internal systems with external assurance workflows and supports work involving NIS2, DORA and ISO 27001.

Private company

Self-funded

Company statement Search funding news
Next, A–ZOuterbounds
← Explore all customers

Orbiq

Orbiq provides trust center and vendor assurance software for business security and compliance workflows. · Security

How they use Nebius Token Factory

Orbiq describes Nebius Token Factory as its primary inference provider after a Mistral outage interrupted a customer demo. It selected the service for EU infrastructure, zero data retention and open-model pricing. Its tasks include language detection, question parsing, compliance answer generation, deep search and questionnaire evaluation.

The implementation assigns Qwen models to different tasks, including Qwen3-14B for language detection and Qwen3-235B variants for answer generation and reasoning. Each AI call passes through a provider chain that checks outage status, retries failed calls and falls back to Mistral. Orbiq reports a 75% reduction in reasoning costs after rebuilding the architecture and changing its model assignments.

Sources

2 sources

How a Demo Outage Forced Us to Rebuild Our Entire AI Inference Architecture

4 Mar 2026 · Orbiq · Web page

Orbiq describes primary inference on Nebius Token Factory, task-specific Qwen models, Mistral fallback and a reported 75% reduction in reasoning costs.

Open original source
How a Demo Outage Forced Us to Rebuild Our Entire AI Inference Architecture4 Mar 2026 · Orbiq · Web pageOrbiq describes primary inference on Nebius Token Factory, task-specific Qwen models, Mistral fallback and a reported 75% reduction in reasoning costs.

Emre Salmanoglu on Orbiq's inference architecture

4 Mar 2026 · LinkedIn · Web page

Emre Salmanoglu describes Orbiq's rebuilt inference architecture with Nebius as primary, multiple-provider fallback and reasoning costs reduced by 75%. The post links Orbiq's technical article.

Open original source
Emre Salmanoglu on Orbiq's inference architecture4 Mar 2026 · LinkedIn · Web pageEmre Salmanoglu describes Orbiq's rebuilt inference architecture with Nebius as primary, multiple-provider fallback and reasoning costs reduced by 75%. The post links Orbiq's technical article.

Company context

orbiqhq.com

Orbiq develops trust center and vendor assurance software for B2B companies. Security, compliance, procurement and legal teams use its platform to share security documentation with buyers and auditors, collect vendor evidence and manage third-party assessments. Its Trust Center Platform provides branded portals with public, restricted and NDA-gated documents. Other features include AI-assisted security questionnaires, vendor evaluations, continuous monitoring and AI search across published compliance content. The platform connects existing internal systems with external assurance workflows and supports work involving NIS2, DORA and ISO 27001.

Private company

Self-funded

Company statement Search funding news