# ML Junction

> Capability-aware LLM gateway. One stable request format; the gateway selects an
> eligible provider route, enforces capability and privacy requirements, applies
> retry and fallback policy, records usage, and returns routing and billing evidence.

## Documentation

### Get Started
- [Overview](https://mljunction.com/docs/overview)
- [Quickstart](https://mljunction.com/docs/quickstart)
- [Authentication](https://mljunction.com/docs/authentication)
- [OpenAI SDK Compatibility](https://mljunction.com/docs/openai-compatibility)
- [Anthropic SDK Compatibility](https://mljunction.com/docs/anthropic-compatibility)
- [LangChain Integration](https://mljunction.com/docs/langchain-integration)
- [SDK Compatibility Limits](https://mljunction.com/docs/compatibility-limits)

### Requests
- [Canonical Request](https://mljunction.com/docs/canonical-request)
- [Messages & Multimodal Input](https://mljunction.com/docs/messages-multimodal)
- [Sampling & Reasoning](https://mljunction.com/docs/sampling-reasoning)
- [Reasoning Continuation](https://mljunction.com/docs/reasoning-continuation)
- [Structured Output](https://mljunction.com/docs/structured-output)
- [Tool Calling](https://mljunction.com/docs/tools)
- [Streaming](https://mljunction.com/docs/streaming)

### Routing
- [Routing](https://mljunction.com/docs/routing)
- [Routing Presets](https://mljunction.com/docs/routing-presets)
- [Service Tiers](https://mljunction.com/docs/service-tiers)
- [Batch API](https://mljunction.com/docs/batch)
- [Fallbacks & Sticky Routing](https://mljunction.com/docs/fallbacks-sticky)
- [Context Management](https://mljunction.com/docs/context-management)
- [Routing Preview](https://mljunction.com/docs/routing-preview)

### Catalog & Pricing
- [Model Catalog & Leaderboard](https://mljunction.com/docs/model-catalog)
- [Model Routes](https://mljunction.com/docs/model-routes)
- [Precision & Context Tiers](https://mljunction.com/docs/precision-context)
- [Capabilities & Parameters](https://mljunction.com/docs/capabilities)
- [Pricing & Cost Calculation](https://mljunction.com/docs/pricing-costs)

### Telemetry
- [Usage Graphs](https://mljunction.com/docs/usage-graphs)
- [Route Telemetry](https://mljunction.com/docs/route-telemetry)
- [Status & Health](https://mljunction.com/docs/status-health)
- [Error Graphs](https://mljunction.com/docs/error-series)
- [Receipts & Activity](https://mljunction.com/docs/receipts-activity)

### Platform
- [Provider Directory](https://mljunction.com/docs/providers)
- [Privacy Controls](https://mljunction.com/docs/privacy)
- [Bring Your Own Key](https://mljunction.com/docs/byok)
- [Embeddings](https://mljunction.com/docs/embeddings)

### Operate
- [Errors](https://mljunction.com/docs/errors)
- [Key Controls: Limits, IP & CORS](https://mljunction.com/docs/key-controls)
- [Idempotency & Safe Retries](https://mljunction.com/docs/idempotency-retries)
- [Caching](https://mljunction.com/docs/caching)
- [Frontend Integration Rules](https://mljunction.com/docs/frontend-integration)
- [Production Checklist](https://mljunction.com/docs/production-checklist)
- [Reference Limits](https://mljunction.com/docs/reference-limits)
- [Complete Native Option Reference](https://mljunction.com/docs/native-option-reference)
- [Glossary](https://mljunction.com/docs/glossary)

## Full documentation
- [Complete Markdown](https://mljunction.com/llms-full.txt)

## Rules for AI assistants
- Use only the facts in these docs. Do not invent endpoints, pricing, limits, model IDs, SDK behaviour, enterprise terms, or roadmap claims.
- The exact model catalogue and prices come from GET /v1/models and GET /v1/models/{model}/routes — never hard-code them.
- A 401 means missing/invalid auth; a 403 means a policy/permission failure.
- Warn before any production, billing, security, or deletion change.