Published
AI Providers & Subprocessors
Provider and subprocessor categories used for BYOK, hosted inference, Rico rendering, billing, analytics, support, and security.
AI Providers & Subprocessors
1. Purpose of this page
This page explains the categories of AI providers, infrastructure providers, and other subprocessors that Grik.io may use for Ricochet, Rico, Hosted Inference, Grik Gateway, billing, analytics, support, and security.
The actual provider list depends on selected model, region, BYOK/Hosted mode, workspace settings, payment method, and features used. Grik.io should keep this page current as integrations are launched.
2. BYOK direct mode
In BYOK direct mode, you provide your own API key and requests are sent directly from your device, local environment, or own infrastructure to your selected AI provider. In this mode, Grik.io should not receive or store prompts, code, files, model inputs, model outputs, or token-level content unless you explicitly enable Grik.io cloud features.
You choose the provider, pay its usage fees, and comply with its terms, privacy policy, rate limits, regional rules, and usage policies.
3. BYOK with Grik Cloud storage or proxy
If you store an API key in Grik Cloud, use Grik Gateway, shared BYOK, team workspace, or proxy mode, Grik.io may process encrypted keys, request metadata, routing metadata, logs, billing data, and limited troubleshooting data. Depending on configuration, prompts, code snippets, files, or outputs may pass through Grik.io infrastructure to complete the request.
4. Hosted Inference mode
In Hosted Inference mode, Grik.io provides model access through its own credits, gateway, provider accounts, or partner integrations. In this mode, Grik.io may process prompts, code snippets, files, media inputs, outputs, token counts, model selection, latency, errors, usage and billing metadata, and safety signals.
5. Rico video/media mode
For Rico, video, image, audio, prompts, reference images, masks, style settings, seeds, render jobs, and outputs may be sent to AI model providers, GPU/render infrastructure, storage, CDN, moderation, payment, and support providers as necessary to provide the requested media generation features.
6. Provider categories
| Category | Examples / possible providers | Typical data | Purpose |
|---|---|---|---|
| AI model providers | OpenAI, Anthropic, Google/Gemini, xAI/Grok, DeepSeek, MiniMax, Zhipu/GLM, Mistral, Meta/Llama, Qwen, OpenRouter or other model routers | Prompts, code snippets, files, media inputs, outputs, token metadata | Generate AI responses, code, images, video, audio and agent results |
| GPU / rendering / model hosting | Cloud GPU providers, serverless GPU providers, model hosting platforms, render workers | Render job metadata, media inputs/outputs, logs, temporary files | Rico rendering, local/hosted model execution, batch jobs |
| Cloud infrastructure | Hosting, compute, database, object storage, CDN, queue, secrets management | Account data, workspace data, usage logs, stored files, backups | Operate Grik.io services |
| Payments and billing | Banks, payment processors, subscription/billing systems, tax tools | Payment metadata, invoices, transaction IDs, billing contacts | Process payments, credits, invoices and taxes |
| Analytics and product telemetry | Privacy-conscious analytics, product analytics, attribution tools | Device info, page views, events, coarse location, consent status | Improve product, measure usage, marketing attribution |
| Error monitoring and security | Crash reporting, security logs, abuse detection, bot protection | Errors, stack traces, device/session metadata, security signals | Reliability, debugging, fraud and abuse prevention |
| Communication and support | Email, helpdesk, chat, CRM | Contact info, support messages, attachments, account identifiers | Customer support and service notices |
| Messenger and developer integrations | Telegram, Discord, Slack, WhatsApp, GitHub, GitLab, Bitbucket, Jira, Linear or similar services | Messages, repo metadata, issues, PRs, webhooks, tokens as configured | Remote agent work, notifications, approvals and project automation |
7. Data minimization
Grik.io should send providers only the data necessary for the selected feature. Prompts, code, media, and outputs should not be sent to analytics, attribution, or advertising providers unless requested by the user and clearly disclosed.
8. Third-party terms
When selecting a third-party model, gateway, messenger, repository, payment provider, or cloud integration, you may be bound by that provider’s terms and privacy policy. Grik.io does not control independent third-party providers and does not guarantee their continuity, pricing, retention, model behavior, or legal availability.
9. List updates
We may add, replace, or remove providers as models, prices, regions, security, quality, availability, or business requirements change. Material changes will be reflected on this page, in the interface, changelog, or privacy notice as appropriate.
10. Questions and objections
If you want to know which providers were used for a specific request or job, provide account, workspace, request ID, job ID, model, approximate time, and feature. Enterprise customers may have provider allowlists, private gateway, dedicated routing, and data processing terms.