ABOUT HOSTLINKU // AI INFRASTRUCTURE & PLATFORM # Dedicated AI Infrastructure. One Unified AI Platform. Hostlinku is an AI infrastructure company combining Dedicated AI Infrastructure with a Unified AI API Gateway. Built as core engineering infrastructure rather than a generic API reseller, Hostlinku delivers flexible inference modes, standard protocol compatibility, and transparent prepaid USD billing for developers, AI agents, startups, and businesses. [Create Free Account](https://api.hostlinku.com/api/auth2/register?lang=en)[View the Model Catalog](https://hostlinku.com/en/models)[Read the Documentation](https://docs.hostlinku.com/en) CORE IDENTITY Dedicated AI Infrastructure CORE PRODUCT Unified AI API Gateway SERVICE OPTIONS Dedicated, Performance, Economy PLATFORM COVERAGE 40 models across 9 vendors 01 // OUR MISSION ## Production AI infrastructure without integration fragmentation Our mission is to bridge production infrastructure discipline with the integration speed modern engineering teams expect. 01 ### The Integration and Compute Divide Engineering teams and AI agent builders routinely face two costly extremes: maintaining separate provider integrations with fragmented credentials and schemas, or absorbing raw compute operations just to keep production throughput predictable. 02 ### Why Hostlinku Exists Hostlinku exists to serve as an infrastructure layer rather than a simple API middleman. We deliver Dedicated AI Infrastructure alongside flexible inference options inside one developer platform with explicit access controls and billing governance. 03 ### Long-Term Vision Our long-term goal is to provide foundational AI infrastructure where developers, AI agents, startups, and enterprises scale cleanly from prototype to production under one API key model, one prepaid USD balance, and consistent runtime behavior. 02 // INFRASTRUCTURE PHILOSOPHY ## Intentional control over capacity, deployment, and workload routing Not every AI workload shares the same operational constraints. Our philosophy centers on structured capacity management, strict key-level controls, and flexible runtime selection without unverified hardware claims. PRINCIPLE 01 ### Access Dedicated deployments without managing your own stack Hostlinku provides Dedicated inference pathways through managed deployments. Your team can integrate models for production applications and agentic workflows without purchasing, building, or operating raw inference infrastructure yourself. PRINCIPLE 02 ### Deployment & Mode Control at the API Key Level Every API key in Hostlinku Console binds explicitly to a chosen Service Mode, granting access to the models available within that mode alongside lifecycle state controls (active, disabled, or revoked), expiration dates, and USD spend caps. PRINCIPLE 03 ### Workload Flexibility Without Rewrites Different workloads do not need the same runtime tier. Separate critical production traffic, real-time apps, and batch pipelines by issuing API keys bound to Dedicated, Performance, or Economy under one uniform API contract. PRINCIPLE 04 ### One Prepaid USD Balance Across All Services Manage inference budgets through a single prepaid USD balance. Charges reflect actual usage across input, output, and supported model caching with micro-USD billing precision and stateless gateway processing. 03 // THREE INFERENCE MODES ## Dedicated at the core, backed by Performance and Economy options Three Service Modes are bound directly at the API key level, enabling teams to standardize on Dedicated infrastructure while using Performance and Economy options wherever workloads warrant them. PRIMARY OFFERING // CORE INFRASTRUCTURE ### Dedicated 7 models active in Dedicated catalog #### Access Dedicated deployments without managing inference infrastructure yourself. Hostlinku delivers Dedicated inference lanes through managed deployments. Your team can integrate models for production applications and agentic workflows without building or operating an inference stack in-house. PURPOSE Production Workload Stability DEPLOYMENT MODEL Managed Dedicated Deployments ACCESS GOVERNANCE Bound per API Key & Mode SUPPORTING OPTION // SPEED26 models ### Performance Performance-optimized inference routing tailored for interactive applications, coding assistants, and latency-sensitive user workflows with broad model coverage across the Unified AI Platform. SUPPORTING OPTION // ECONOMY21 models ### Economy Cost-efficient inference option designed to minimize per-token cost across background pipelines, automated evaluations, batch classification, and high-volume workloads using the same prepaid balance. 04 // UNIFIED AI PLATFORM ## One API to access multiple AI model families Integrate multiple model families through a single gateway and consistent API interface. Select models for reasoning, coding, and multimodal workloads across 40 models and 9 vendors without building separate integrations. Each Hostlinku API key binds to one Service Mode and accesses all models available within that selected mode. [Browse the Model Catalog](https://hostlinku.com/en/models)[SDK Integration Docs](https://docs.hostlinku.com/en) ### OpenAI Chat Completions `POST /v1/chat/completions` Drop-in compatibility with standard OpenAI clients and SDKs for chat, tool calling, SSE streaming, and JSON structured outputs. ### Anthropic Messages `POST /v1/messages` Supports both Messages and count\_tokens endpoints via x-api-key or Authorization Bearer headers, including native prompt caching. ### OpenAI Responses `POST /v1/responses` Runs statelessly (store=false) at the gateway boundary for tool invocation and structured outputs without server-side conversation state storage. ### Dynamic Model Catalog `GET /v1/models` Returns the live active model catalog automatically filtered by the calling API key's bound Service Mode. 05 // TECHNICAL ARCHITECTURE ## Public execution architecture from gateway edge to Service Mode This interactive schematic illustrates Hostlinku's public execution flow: API key authentication, atomic prepaid balance reservation, Service Mode pool dispatch, and micro-USD settlement without exposing private operational internals. PUBLIC INFRASTRUCTURE EXECUTION FLOW Select a Service Mode and API dialect to inspect how requests move through the Hostlinku gateway. MODE 01DedicatedPrimary Infrastructure 7 models available Dedicated AI Infrastructure provisioned for sustained production applications, autonomous agents, and core business workflows requiring consistent execution posture. WORKLOAD FIT Production AI agents, recurring core services, operational automation. CAPACITY MODEL Managed capacity aligned at the API key and deployment level. API DIALECT COMPATIBILITY POST/v1/chat/completions Authorization: Bearer sk-... SSE Streaming, Tool Calling, JSON Schema Full compatibility with standard OpenAI Chat Completions SDKs and clients without changing application logic. 01TLS + SHA-256 ### Edge & Key Auth Validates TLS ingress, SHA-256 API key hash, active status and expiry checks, and the key's bound Service Mode (Dedicated). 02Atomic USD Ledger ### Spend Cap & Reserve Enforces configurable spend limits (daily, weekly, monthly, or lifetime) and reserves available prepaid USD atomically prior to execution. 03Dedicated / Perf / Economy ### Service Mode Dispatch Dedicated Deployment Lane processes POST /v1/chat/completions across 7 active models in this mode. 04Micro-USD (1e-8) ### Stream, Meter & Settle Streams SSE output to the client, logs token telemetry without default payload storage, settles micro-USD cost, or releases reservations automatically on pre-output errors. BASE URL`https://api.hostlinku.com/v1` [View the Model Catalog](https://hostlinku.com/en/models)[View System Status](https://status.hostlinku.com/) 06 // SECURITY & RELIABILITY ## Verified safeguards across access control, data privacy, and billing integrity [Security Policy](https://docs.hostlinku.com/en/legals/security)[Privacy Policy](https://docs.hostlinku.com/en/legals/privacy-policy)[System Status](https://status.hostlinku.com/) SEC-01 ### API Key Security & Controls - Raw API key secrets are revealed once upon creation and persisted exclusively as SHA-256 cryptographic hashes. - Each key supports independent lifecycle states (active, disabled, revoked), optional expiration timestamps, Service Mode binding, and USD spend caps (daily, weekly, monthly, or lifetime). SEC-02 ### Authentication, RBAC & Surface Separation - Console access uses Keycloak OIDC Authorization Code with PKCE (S256), email OTP verification, optional TOTP 2FA or passkeys, Cloudflare Turnstile checks, and HttpOnly, Secure, SameSite=Lax session cookies. - Teams use Organization RBAC (admin and member roles). Browser session cookies cannot invoke /v1/\* inference endpoints, and API keys are rejected on /api/\* account and billing management routes. SEC-03 ### Stateless Execution & Payload Privacy - The API gateway executes statelessly by default without storing prompt or response bodies, and customer payloads are not used to train Hostlinku models; upstream model provider data handling follows each provider's retention policy. - Prompt and response body logging is disabled by default. When diagnostics are explicitly enabled, credential patterns are redacted prior to time-bounded retention, and sensitive actions are written to the security audit log. SEC-04 ### Unified Prepaid USD Balance & Billing Integrity - Manage inference spending from one prepaid USD balance with zero mandatory monthly or per-seat subscription fees. Costs are metered from actual usage across input, output, and supported model caching on an atomic micro-USD ledger (1e-8 USD). - Requests that fail before generating output release reserved funds automatically with zero charge. Teams can inspect token and latency logs, export CSV records, and review automated health probes on the Status page. GET STARTED ## Ready to accelerate your AI deployments? Create an account in seconds, test models in the Console Playground, or connect production workloads to Hostlinku's Dedicated AI Infrastructure and Unified AI Platform today. [Get Started for Free Now](https://api.hostlinku.com/api/auth2/register?lang=en)[View the Model Catalog](https://hostlinku.com/en/models)[Read the Documentation](https://docs.hostlinku.com/en)[Contact the Team](https://hostlinku.com/en/contact)