# PRYSM

> One API, every model, always the best one. PRYSM is a universal AI routing engine — an OpenAI drop-in that routes every prompt to the best-value model across 21 models and 9 providers.

## Docs

- [Validate BRAIN.md](https://docs.prysm1.com/api-reference/brain-validate.md): Validate a BRAIN.md config — raw text or an object — against the model catalog and schema.
- [Create chat completion](https://docs.prysm1.com/api-reference/chat-completions.md): Route a prompt to the best model and return an OpenAI-compatible completion enriched with a prysm block.
- [Code](https://docs.prysm1.com/api-reference/code.md): Generate code with a self-correcting, multi-model loop — a coder writes it, a separate critic reviews it, the coder repairs against the feedback — and get a PrysmProof v2.
- [Compliance preview](https://docs.prysm1.com/api-reference/compliance-preview.md): Dry-run the Policy-as-Code compliance gate — classify a prompt and see which models a policy allows vs excludes, the Compliance Cost Premium, and a sample attestation. No model is called.
- [Health](https://docs.prysm1.com/api-reference/health.md): Service health check with model count, provider status, and total requests served.
- [API reference](https://docs.prysm1.com/api-reference/introduction.md): Base URL, authentication, content types, the response envelope, and error format for the PRYSM REST API.
- [List models](https://docs.prysm1.com/api-reference/models.md): Retrieve the full catalog of 21 models with pricing, tier, context window, and strengths.
- [Orchestrate](https://docs.prysm1.com/api-reference/orchestrate.md): Run a prompt across several models — cascade, ensemble, decompose, or debate — and get one synthesized answer with a PrysmProof v2.
- [Verify proof](https://docs.prysm1.com/api-reference/proof.md): Verify a PrysmProof receipt by request ID — public, so anyone can independently confirm what ran.
- [Preview routing](https://docs.prysm1.com/api-reference/route.md): Dry-run the router: see which model PRYSM would pick and the estimated cost — without calling the model.
- [Get usage](https://docs.prysm1.com/api-reference/usage.md): Usage statistics for the authenticated key — totals plus breakdowns by model, provider, and mode.
- [Authentication](https://docs.prysm1.com/authentication.md): How to authenticate with the PRYSM API: keys, the base URL, and environment variables.
- [AgentGuard](https://docs.prysm1.com/concepts/agentguard.md): Cost guardrails that keep autonomous agents from overspending — per-request caps, monthly budgets, and model blocks.
- [BRAIN.md](https://docs.prysm1.com/concepts/brain-md.md): A declarative, version-controlled routing config — the .cursorrules of model routing.
- [How Code Mode works](https://docs.prysm1.com/concepts/code-mode.md): A coder model writes the solution, a separate critic reviews it, and the coder repairs against that feedback until it passes — self-correcting, multi-model code generation with a PrysmProof.
- [Compliance routing](https://docs.prysm1.com/concepts/compliance.md): Policy-as-Code: load your approved providers and PRYSM routes every request only to models that satisfy your jurisdiction, framework, certification, and data-residency rules — with an auditable attestation of where each request could go.
- [How orchestration works](https://docs.prysm1.com/concepts/orchestration.md): Routing picks the one best model. Orchestration crosses several — to cut cost or to raise robustness beyond any single model — and proves how it did it.
- [PrysmProof](https://docs.prysm1.com/concepts/prysmproof.md): A tamper-evident SHA-256 receipt on every response — proving which model ran, in which mode, on which input and output.
- [How routing works](https://docs.prysm1.com/concepts/routing.md): PRYSM classifies the intent of each prompt and selects the best-value model — in one of three modes.
- [Introduction](https://docs.prysm1.com/introduction.md): One API, every model, always the best one. PRYSM routes every prompt to the best-value model across 21 models and 9 providers — without changing your code.
- [Model catalog](https://docs.prysm1.com/models.md): The 21 models PRYSM routes across, with live pricing, context windows, and strengths — grouped by tier.
- [Quickstart](https://docs.prysm1.com/quickstart.md): Make your first routed request in under five minutes.
- [CLI](https://docs.prysm1.com/sdks/cli.md): The prysm command — preview routing, list models, validate BRAIN.md, check usage, and chat from your terminal.
- [MCP server](https://docs.prysm1.com/sdks/mcp.md): @prysmai/mcp — give Claude Desktop, Cursor, Windsurf, or any MCP client cost-aware routing as a native tool.
- [Node SDK](https://docs.prysm1.com/sdks/node.md): @prysmai/sdk — a fully-typed OpenAI drop-in for Node/TypeScript with routing, cost, proof, and BRAIN.md auto-discovery.
- [Python SDK](https://docs.prysm1.com/sdks/python.md): prysm1 — an OpenAI drop-in for Python with routing, cost, proof, and BRAIN.md auto-discovery.