Skip to main content
Leonex
Provider-neutral AI infrastructure

The infrastructure layer for production AI.

Build against one dependable gateway. Test models, manage credentials, observe every request, and keep control as your AI stack evolves.

  • Project-scoped keys
  • Atomic idempotency
  • Normalized contracts
gateway.pipeline
Leonex request pipelineA request passes through validation and policy evaluation, resolves through a provider-neutral interface, and returns a normalized response.RequestValidatePolicyResolveResponserequest_id + correlation_id + context
DeterministicObservableProvider-neutral

A calmer control plane

Everything around the model, designed as infrastructure.

Leonex gives engineering teams a consistent boundary for credentials, policies, observability, and future provider adapters.

One stable gateway

A provider-neutral request contract keeps application code separate from model-specific infrastructure.

Scoped credentials

Organize live and test keys by project, with expiration, rotation, revocation, and one-time secret display.

Safe experimentation

Move from model exploration to production through explicit environments and repeatable request shapes.

Request visibility

Request and correlation identifiers make each operation traceable across the gateway pipeline.

Usage foundations

Clean recording interfaces prepare the platform for trustworthy usage and cost reporting as adapters arrive.

Enterprise boundaries

Organizations, memberships, roles, projects, and audit events provide durable ownership boundaries.

Interactive pipeline

How Leonex turns a request into controlled infrastructure.

Select a stage to inspect the gateway boundary. The flow is explicit, observable, and adapter-independent.

STAGE 01

Receive

Leonex accepts a normalized request envelope and creates immutable request and correlation identifiers.

Outcome

One entry point for every supported operation.

Model layer

Choose capabilities without hard-wiring your product to a provider.

Leonex models the operations your application needs. Provider adapters can evolve behind that stable boundary while availability stays explicit.

Availability is never simulated

Only verified adapters and models appear as selectable. Current availability is published in the platform documentation.

Capability contractsProvider-neutral

text/generation

Normalized prompts and messages

vector/embedding

Consistent vector operations

response/structured

Schema-aware output contracts

transport/streaming

Incremental response interfaces

Example request
test
curl -X POST /gateway/v1/inference \
  -H "Authorization: Bearer lnx_test_..." \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: request-123" \
  -d '{
    "operation": "generate",
    "model": "your-model-alias",
    "input": "Summarize this incident."
  }'

Developer experience

A clean contract at the edge of a complicated ecosystem.

Use the same request shape from an early prototype through a production service. Infrastructure concerns stay visible without leaking into product code.

  • One request envelope
  • Stable error contract
  • Request and correlation IDs
  • Streaming-ready interfaces

Enterprise foundations

Control that grows with your organization.

Leonex is structured around ownership, isolation, and traceability so larger teams do not have to retrofit governance later.

Explore enterprise readiness

Durable ownership

Organizations and projects create clear boundaries for credentials, usage, and future billing.

Least privilege

Roles, permissions, scopes, expiration, rotation, and revocation are designed into the identity model.

Audit-ready events

Security-sensitive actions produce structured audit and domain events for future retention policies.

Scale-aware architecture

Stateless services, clean ports, and asynchronous recording boundaries prepare the system for horizontal scale.

Pricing preview

A commercial model designed to scale with usage.

Plan structure is shown for product context only. Final limits, rates, and commercial terms will be published before general availability.

Build

Developer access

For individual developers validating an AI product.

  • Test and live project boundaries
  • Scoped API key lifecycle
  • Gateway request visibility

Scale

Designed for teams

Usage-based

For teams operating customer-facing AI workloads.

  • Organization workspaces
  • Operational usage views
  • Expanded policy controls

Enterprise

Custom agreement

For organizations with advanced control requirements.

  • Advanced access governance
  • Audit retention options
  • Deployment and support planning

FAQ

Questions, answered directly.

Clear platform boundaries matter. Here is what Leonex does, and what remains deliberately explicit.

What problem does Leonex solve?

Leonex provides a stable infrastructure boundary between an application and its AI capabilities. It centralizes request contracts, project credentials, policy evaluation, idempotency, observability, and future provider resolution.

Does Leonex lock us into a specific model provider?

No. The gateway domain is provider-neutral by design. Provider adapters implement explicit interfaces, so application code depends on Leonex capability contracts rather than a vendor-specific SDK.

Which providers and models are available?

Leonex does not advertise planned integrations as live. Verified adapters and their supported models will be listed in the platform documentation when they are production-ready.

How are API key secrets handled?

Project API keys use environment-specific prefixes, are hashed at rest, and reveal their plaintext secret only once. The lifecycle supports expiration, rotation, revocation, scopes, and optimistic concurrency.

Is the platform designed for enterprise teams?

Yes. Organizations, multi-organization membership, roles, permissions, projects, sessions, and audit events establish the ownership boundaries needed for larger customers and future enterprise controls.

Is the pricing on this page final?

No. The pricing area is a non-functional product preview. Final rates, limits, included capabilities, and commercial terms will be published before general availability.

Build on a stable boundary

Your AI stack should stay flexible as requirements change.

Start with the architecture, inspect the request pipeline, and build against interfaces designed for production constraints.