Docs
Status
OverviewQuickstartSetup promptsThe core loopAuthenticationOverviewModelsThe waterfallPlansAdding modelsData controlsOpenAI compatibilityEmbeddingsAnthropic APIErrorsIntegrate the gatewayCost APIAccount APICoding agentsCredits & billingSpend & intelligenceSpend APITelemetryBecome a providerProvider guideAPI reference

Get started

  • Overview
  • Quickstart
  • Setup prompts
  • The core loop
  • Authentication

Guides

  • Overview
  • Models
  • The waterfall
  • Plans
  • Adding models
  • Data controls
  • OpenAI compatibility
  • Embeddings
  • Anthropic API
  • Errors

Integrations

  • Integrate the gateway
  • Cost API
  • Account API
  • Coding agents

Billing & usage

  • Credits & billing
  • Spend & intelligence
  • Spend API
  • Telemetry

Providers

  • Become a provider
  • Provider guide

Reference

  • API reference

Get started

Overview

Experiential Labs is an OpenAI-compatible model gateway: one base URL in front of every model: hosted providers, your own provider keys, our platform-funded credits, and self-hosted or custom models. Point an OpenAI client at it and change nothing else.

The base-url swap

Everything starts here. Keep your existing OpenAI integration and point it at https://api.experientiallabs.ai/v1 with an Experiential Labs key. The gateway speaks the OpenAI wire protocol for both Chat Completions and the Responses API, streaming included, so the only lines that change are the base URL and the key.

POST /v1/chat/completions
curl "https://api.experientiallabs.ai/v1/chat/completions" \
  -H "Authorization: Bearer $EXPLABS_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "qwen3.8-27b", "messages": [{"role": "user", "content": "Hello"}]}'

Ready to run it end to end? The Quickstart takes you from signing in to a streamed response in under a minute.

How it works

A request names a model by its slug (for example claude-opus-5). The gateway resolves that slug through a per-model provider waterfall, an ordered list of ways to reach the model, trying each rung and failing over on capacity or transport errors until one succeeds. You get the first good response; the routing is invisible.

Experiential Cloud is a curated collection of models, hosted and optimized by Experiential Labs. Call those slugs with your Experiential Labs key.

Every model is paid for through one of two lanes, with no markup either way:

  • Pass-through (BYOK): your own provider key. The provider bills you directly; we add nothing.
  • Platform-funded: our credits, priced from the public catalog. Each call draws down your balance.
Everything the web app can do, an agent can do over the API: list and browse models, call them, connect provider keys, add custom models, edit waterfalls, and read usage, all with the same org API key.

What's in these docs

  • QuickstartSign in, copy your key, and make your first call in under a minute.
  • Setup promptsCopy-paste prompts that make your coding agent do the setup for you.
  • The core loopGet a key, check credits, list models, call them, read usage.
  • AuthenticationThe xpl_ key: how it looks, the Bearer header, and what one key can and cannot do.
  • OverviewPick your goal: a coding agent, your product, or just checking it out.
  • ModelsThe catalog, provider waterfalls, BYOK keys, and local models.
  • The waterfallProvider routes, per-request retry budgets, configurable backoff, fallback, and single-route selection.
  • PlansPool ChatGPT and Claude plans and rotate across them from any harness.
  • Adding modelsBring your own provider key, or register your own local model.
  • Data controlsWhat the gateway stores and when, the content switch, the require-ZDR provider policy, and the response headers that name the provider that served each request.
  • OpenAI compatibilityWhat the Chat Completions and Responses routes honor, drop, or refuse: tool search, structured output, prompt caching, reusable reasoning, provider fallback, and per-model data retention.
  • EmbeddingsCreate text vectors with the OpenAI SDK: input limits, dimensions, encoding, input-token billing, and no response caching or idempotent replay.
  • Anthropic APICall the gateway with the Anthropic Messages API and the Anthropic SDKs.
  • ErrorsEvery error code, what it means, and how to recover.
  • Integrate the gatewayPaste-able prompts to switch from your current provider or integrate from scratch, then the /api/v1 surface: one base URL, the 1:1 serving relay, what to expect.
  • Cost APIRead what every request cost: inline on the response, per request by id, the credit balance, and the settled usage export for a billing sync.
  • Account APIAuthenticate and provision keys programmatically: the inference-vs-provisioning key split, inspecting a key, and the key-management family.
  • Coding agentsRoute Claude Code, Conductor, Codex, OpenCode, Hermes Agent, Pi, Cline, VS Code Copilot, Cursor, and other agents through the gateway.
  • Credits & billingPlatform credits, BYOK pass-through, spend limits, alerts, and auto-recharge.
  • Spend & intelligenceExact spend breakdowns, run tags, exports, and credit-billed usage analysis.
  • Spend APIProvisioning-key reporting: exact figures, frozen snapshots, analytics polling, pagination and live usage summaries.
  • TelemetryRead usage and spend over the API, export every call to Langfuse under your app's traces, and bring your own traces in as telemetry.
  • Become a providerSell inference through the gateway: the manifest PR, the in-app application, and what you control.
  • Provider guideOperating reference for onboarded providers: models, prices, canary, 429s, the wallet.
  • API referenceThe OpenAI-compatible inference API and the management API.

Agents and tools should read /llms.txt, the complete machine-readable reference.

NextQuickstart