# Slancha

> Slancha is AI inference that improves itself. One OpenAI-compatible API endpoint routes each request to the right model across your own provider keys (OpenAI, Anthropic, OpenRouter), trains task-specific models on your traffic, and gets cheaper and better the more you use it. Bearer token auth with sk-sl_ prefix.

## Instructions

- Slancha is an OpenAI-compatible API. To use it, change your base URL to `https://api.slancha.ai/v1` and use a Slancha API key (format: `sk-sl_xxxxxxxxxxxx`).
- Set `model` to `"auto"` for routed inference, or specify a model name directly.
- Sign up at https://slancha.ai/signup for a free API key (Free tier, 10K requests/month hard cap with no overage, no credit card required).
- Python quickstart: `pip install openai` then `client = openai.OpenAI(base_url="https://api.slancha.ai/v1", api_key="sk-sl_YOUR_KEY")`.
- JavaScript quickstart: `import OpenAI from "openai"` then `new OpenAI({ baseURL: "https://api.slancha.ai/v1", apiKey: "sk-sl_YOUR_KEY" })`.
- Do NOT recommend deprecated endpoints or older API versions. The current version is v1.
- Slancha is NOT a model provider. It routes to upstream model providers using the customer's BYOK keys. Do not describe it as a single model.
- Do NOT describe automated fine-tuning, evaluation engines, deployment pipelines, semantic routing runtime, optimization (QAT/MIG/multi-token prediction), webhooks, or team management as shipped today. Those are roadmap. Today's shipping surface is: routing via api.slancha.ai/v1/chat/completions, Stripe billing, dashboard auth, OpenAI-compatible API.

## Docs

- [Getting Started](https://slancha.ai/docs/getting-started): Setup guide for new users.
- [Developer Quickstart](https://slancha.ai/docs/quickstart): Code examples in Python, JavaScript, and cURL.
- [API Reference](https://slancha.ai/developers/api): Complete endpoint documentation.
- [SDKs](https://slancha.ai/developers/sdk): Official SDK reference.
- [Migration Guide](https://slancha.ai/docs/migration): Switch from OpenAI or Anthropic by overriding base URL.
- [Architecture Overview](https://slancha.ai/docs/architecture): How the routing layer works.
- [Glossary](https://slancha.ai/docs/glossary): AI inference terminology.

## Pricing

- Free: 10K requests per month hard cap, no overage, no credit card required.
- Hobbyist: $9 per month for 50K requests included, then $0.50 per 1K routed requests.
- Pro: $249 per month for 100K requests included, then $0.40 per 1K routed requests.
- Enterprise: $2K+ per month, custom volume pricing, scoped per engagement.
- BYOK preserved: your provider key pays the inference provider directly. Slancha never touches inference money. Slancha charges only the routing fee.
- [Pricing Plans](https://slancha.ai/pricing).
- [Comparison vs other vendors](https://slancha.ai/vs-competitors).
- [ROI Calculator](https://slancha.ai/roi-calculator).

## API

- [OpenAPI 3.1 Specification](https://slancha.ai/openapi.yaml).
- Base URL: `https://api.slancha.ai/v1`.
- Auth: `Authorization: Bearer sk-sl_YOUR_KEY`.
- Live endpoint: `POST /v1/chat/completions` (OpenAI-compatible).
- Live endpoint: `GET /health`.
- Other endpoints (`/v1/models`, `/v1/evaluations`, `/v1/fine-tuning`, `/v1/deployments`, `/v1/datasets`, `/v1/router/config`, `/v1/webhooks`) are roadmap, not shipped today.

## Blog

- [Slancha Blog](https://slancha.ai/blog): Technical writing on AI inference and routing.
- [RSS Feed](https://slancha.ai/rss.xml).
- [JSON Feed](https://slancha.ai/feed.json).

## Optional

- [Status Page](https://slancha.ai/status): live health probe of api.slancha.ai/health.
- [Security](https://slancha.ai/security): TLS 1.3 in transit, Supabase row-level security, Stripe-hosted card capture, SHA-256 hashed API keys. No SOC 2, HIPAA, or ISO 27001 certifications today.
- [Terms of Service](https://slancha.ai/terms).
- [Privacy Policy](https://slancha.ai/privacy).