Terms of Service

RouteCortex provides inference infrastructure for open-weight AI models to upstream platforms (aggregators, developer tooling). These terms govern the use of our API endpoints at api.routecortex.io.

Nature of service

We operate as an inference-layer provider. We do not create or curate the models we serve — all models are open-weight releases sourced from public model repositories (Hugging Face and equivalent). Model quality, alignment, and behavior are properties of the upstream model authors.

Acceptable use

Access to our inference endpoints is contracted through upstream platform partners. Each partner is responsible for enforcing their own acceptable-use policy against their end users. We reserve the right to refuse or terminate routing on the following categories:

• Content involving minors in a sexual context
• Coordinated illegal activity (fraud, malware distribution, targeted harassment campaigns)
• Traffic that violates the upstream model’s license terms

Beyond these categories, we do not moderate or filter inference content. We are the raw compute layer.

Availability

We target 99% availability across our routed models but do not guarantee uptime for any specific model or capacity level. Scheduled maintenance, upstream GPU cloud disruptions, or preemption of interruptible instances may cause temporary outages.

Aggregators integrating our endpoints should implement failover to alternative providers for their user-facing SLA needs.

Pricing and payment

Per-token pricing is set per model and disclosed to integrating aggregators at onboarding. Payment is processed monthly on invoice, settled in USDC on Base network or equivalent stablecoin rail agreed with the partner.

Content and IP

We do not claim ownership over prompts routed through our infrastructure or completions generated by the models we serve. Copyright and licensing of model outputs is governed by the underlying model’s license and by the aggregator’s own terms with its end users.

Liability

RouteCortex provides inference on a best-effort basis without warranties of fitness for a particular purpose. Aggregators are responsible for verifying model outputs are suitable for their use case before serving to end users. Our aggregate liability to any integrating partner is limited to the amounts paid by that partner in the trailing three calendar months.

Changes

We may update these terms with 30 days notice to integrating partners via the contact email on record. Continued use of the API after the notice period constitutes acceptance.

Contact

Written correspondence: contact@routecortex.io