Experiential Labs · 2026-08-28 · notable
Experiential — an open-source model gateway that takes no token markup
Experiential is an Apache-2.0 model gateway written in Rust that puts hosted, open-source, local and custom models behind one OpenAI-compatible API. The team charges provider prices with no markup and launched it on Show HN.
Experiential puts self-hosted, frontier and open-source models behind one OpenAI-compatible API, and charges no markup on top of provider prices.
Key specs
| GitHub stars | 719 |
|---|---|
| Added latency (byok) | under 1 ms |
What is it?
One control plane across closed, open-source, local and custom models is what Experiential offers: a Rust gateway, released under Apache-2.0, that speaks the OpenAI and Anthropic Messages APIs and hands requests to whichever backend you point it at. Experiential Labs launched it on Show HN on August 28, where it drew 213 points and 46 comments. It ships as a Python package you install and run locally, or as a hosted endpoint at api.experientiallabs.ai.
How does it work?
The gateway absorbs the differences between providers — streaming formats, tool calls, model parameters, rate limits and error behaviour — so callers see one shape. The team says it adds under 1 ms to a bring-your-own-key request and under 2 ms when Experiential supplies the provider key, and that a codex agent refreshes its 1,000-plus model list daily by opening a pull request. Failover routes the same model across more than one provider, and per-user and per-agent controls track spend.
Why does it matter?
Routing is the part of an LLM stack teams usually rent, and the common gateways charge a percentage of every token that passes through. Experiential's pitch is that a routing layer does not justify that cut: run it yourself, or use the hosted version and pay provider prices. Experiential Labs also offers an opt-in path that turns your own OpenTelemetry traces into a routing policy — it mines representative tasks from them, simulates rollouts for different models, scores those with an LLM judge, and fits a nearest-neighbour classifier over a prompt embedding to pick a model per request.
Who is it for?
teams running several model providers behind one API
Try it
pip install experiential && exp