One endpoint for multiple model sources
Route hosted providers, your own keys, and your own GPUs through a single OpenAI-compatible endpoint so agents can use one API key across model sources.
Experiential Labs
Experiential Labs is an open source AI gateway that routes requests across hosted providers, your own keys, and your own GPUs through one OpenAI-compatible endpoint. It adds access control, attribution, budgets, and an intelligence layer for model routing and caching.
Open source gateway for routed AI access
Experiential Labs is an open source AI gateway that puts multiple model providers behind one OpenAI-compatible endpoint. The product is built around a single API key and routing layer that can work with hosted providers, customer-owned keys, and customer-owned GPUs.
The site positions the gateway as a way to control routing, access, and attribution for agent traffic while keeping the provider's price on routed tokens. It also adds an intelligence layer that can suggest model changes, caching opportunities, and model training or optimization based on observed usage, with the option to self-host the whole gateway on your own infrastructure.
Core capabilities
Access control and spend controls
Set budgets, hard caps, model allowlists, and key scopes at the org, team, person, agent, or tool level so requests are constrained at the gateway.
Request-level attribution and reporting
Attribute every request with route, fallback, tokens, and cost, then review usage by agent, person, model, or day in the console and usage API.
Traffic-aware routing suggestions
Use the intelligence layer to recommend cheaper or faster model routes, identify cache opportunities, and support per-prompt optimization based on observed traffic.
Prompt caching for repeated work
Turn on prompt caching for repeated tokens and keep the same prompts and workflow while reducing the effective token cost of repeated content.
Hosted or self-hosted deployment
Run the gateway on your own infrastructure or use the hosted service, with the core available as open source and self-hosting documented on the site.
Where it fits
Unify model access across an application stack
Point agents, apps, or internal tools at a single base URL so model selection, provider failover, and cost attribution happen at the gateway instead of in each application.
Control spend and access for teams
Use org policies, caps, and allowlists to keep a team, agent, or key within a planned budget while still letting requests route across multiple providers.
Reduce cost on recurring workloads
Route repeated prompts through caching and let the intelligence layer surface cheaper routes or lower-effort settings when traffic patterns support it.
Run the gateway in your own environment
Self-host the gateway when you want the endpoint, budgets, routing, and usage API on your own infrastructure rather than a managed deployment.
Deploy a model tuned to your workflow
Use the case-study pattern to train or distill a model on your own workflow, then serve it behind the same endpoint as the rest of your model traffic.
Pros and Cons
Pros
- Combines hosted providers, BYOK, and local or self-hosted models behind one endpoint.
- Uses provider list price for routed tokens with 0% markup on routed tokens.
- Adds budgets, hard caps, model allowlists, and request attribution at the gateway.
- Offers an OpenAI-compatible /v1 endpoint and streaming support.
- Can be self-hosted on your own infrastructure with open source core available.
Cons
- The intelligence layer is not free on every plan; the pricing page says it is included in Pro and Enterprise and billed in credits.
- Some enterprise capabilities are described as planned or available on request, including managed upgrades and backups, short-lived credentials, and some security review or support arrangements.
FAQ
How do you set it up?
The gateway exposes an OpenAI-compatible /v1 endpoint. The pricing page says the self-hosted gateway can be run on your own infrastructure with `uvx --from experiential exp run`, and the site shows examples for standard chat-completions requests.
Who is it for?
It is positioned for teams that want one endpoint for hosted providers, BYOK, and local or self-hosted models, plus routing, budgets, access control, and attribution. The pricing page also describes enterprise controls such as SSO, SCIM, advanced RBAC, private networking, and data residency.
What does the intelligence layer do?
The gateway routes requests through the cheapest healthy deployment that can serve a model, and the intelligence layer can also suggest model switches, caching opportunities, and per-prompt optimization based on observed traffic.
How does pricing work?
The site says routing uses the provider's list price with 0% markup on routed tokens. Credits are used across routed tokens and intelligence features, and requests stop at caps when credits run out.
Can you use your own keys or models?
Yes. The pricing page says BYOK and local models are included on Pro and Enterprise, and the home page says your own fine-tuned and self-hosted models can sit behind the same key.
Quick Facts
- Category
- AI gateway / developer tool
- Platform
- Web-based service with self-hosting option
- Primary users
- Teams and developers routing agent traffic across models
- Source domain
- experientiallabs.ai
- API style
- OpenAI-compatible /v1 endpoint
- Pricing signal
- Free plan, Pro plan, and Enterprise plan
Alternative a Experiential Labs
AakarDev AI
AakarDev AI helps teams manage AI provider access, project-level setups, logs, and analytics from one dashboard. It supports BYOK workflows and lists providers including OpenAI, Google Gemini, Anthropic, Groq, Mistral AI, and Perplexity AI.
ByteAsk
ByteAsk is a terminal-first AI coding agent for C and C++ that edits repositories and verifies changes with the real compiler, debugger, sanitizers, and tests before showing a diff. It offers a free tier plus paid plans, with editor connectors and zero-retention handling described in the source.
CreateOS Sandbox
CreateOS Sandbox is an isolated compute environment for running code and agent workloads inside Firecracker micro-VMs. It is designed for workflows that need machine-level isolation, private networking between sandboxes, and programmatic control through SDK, CLI, or MCP.
hob
hob is an independent workspace for coding agents that keeps agent sessions, terminals, history, and follow-up work organized around the tools and providers you already use. It is aimed at developers who want local control over routing, history, and workspace structure rather than a bundled model stack.
Ably Chat
Ably Chat is a chat API platform for building custom realtime chat applications. It supports room-based messaging, typing indicators, presence, reactions, and message updates, with usage-based pricing options for different deployment stages.
Manta AI
Manta AI is an autonomous web app testing tool for teams that want to map application behavior, catch regressions, and generate tests without writing scripts or maintaining selectors. It works from a URL and supports plain-English test flows, run results with screenshots, and scheduled or deployment-triggered checks.