Experiential
3.2k

Backed by Y Combinator

Every AI model. One key.
Zero markup.

The open source AI gateway. Hosted providers, your own keys, and your own GPUs behind one endpoint, at the provider's price.

POST api.experientiallabs.ai/v1
Book a call
one key

POSTmodel=

Pay the provider's price.

0%Markup on routed tokens

The gateway is free and open source. We earn on hosted inference and on Pro. See pricing

Make your AI better, faster, cheaper.

The gateway routes. The intelligence layer improves what flows through it.

How the intelligence layer works →

The right model

We watch your traffic and tell you when a switch wins, including new models the day they ship. Turnkey per-prompt optimization, if you want it.

Caching

We catch opportunities to improve cache hit rate, and show what they would save. Repeated tokens come back at 90% off when you turn it on.

A model you own

Fine-tuned on your traffic, proved before it serves, and reached through the same endpoint. You turn it on; it never happens on its own.

Case studies

A model you own, trained on your workflow, doing the job you rent a frontier model for at a tenth of the price and several times the speed.

Computer use agent · July 2026−97%cost, same tasksA 9B model distilled from a teacher working 1,000 tasks on Mac VMs, then scored in closed-loop simulation: 50% more tasks completed than the untrained 9B, 7× faster than Opus 4.8.Read the story →Claims research · August 2026+10.9%verdict accuracyA 4B model trained with GRPO on the public AVeriTeC benchmark and run on 500 held-out claims with the same tools and turn limit as Haiku 4.5: 9.4× faster, at 90% lower cost.Read the story →

Every request and every dollar,
in one console.

Catalog, usage, and limits for the whole org, by agent, person, model, or day.

Open the console →
Acme CorpAttribution · Augustapi.experientiallabs.ai
modelroutectx30d tokensspend
fable-5anthropic → bedrock1M38.1M$1,942
gpt-5.6openai → azure400K24.6M$1,118
gpt-5.6-solopenai400K14.3M$486
gemini-3.7-flashgoogle2M11.2M$96
haiku-4.5anthropic200K9.8M$88
glm-5.3zhipuyour key200K6.1M$34
qwen3.8-27byour gpuslocal128K31.4M$0.00

Spend by agent, person, model, or day.

Point your agent here.

Change the base URL and keep your code. Every model answers from one endpoint, and streams pass through token by token. The last tab is a prompt: hand it to Claude Code, Cursor or Codex and it runs the setup for you.

DocsGitHub
curl https://api.experientiallabs.ai/v1/chat/completions \  -H "Authorization: Bearer $EXPLABS_API_KEY" \  -H "Content-Type: application/json" \  -d '{ "model": "fable-5", "messages": [{ "role": "user", "content": "Hello" }] }'

Give your whole company every model.

One API and one login, scoped by team or key. We handle key management, new models land on day zero, and your own fine-tuned and self-hosted models sit behind the same key.

Models
  • OpenAI
  • Anthropic
  • Gemini
  • Meta
  • Qwen
  • GLM
  • Kimi
  • DeepSeek
  • Mistral
  • Your APIs
  • Local
Inference providers
  • Bedrock
  • Azure AI Foundry
  • Fireworks
  • Modal
  • Vertex
  • OpenRouter
  • DoubleWord
  • Tencent Cloud

Know what your AI spend will be.

One key per person, agent, and tool. Caps set once by an admin, enforced on every request, so the bill is the number you planned.

  • Keys with caps, by day, week, or month.
  • Roles, model allowlists, local-only scopes.
  • Self-host it, or use the hosted gateway.
Enterprise plans →
Acme CorpKeys · 23 active · 1 at cap
support-agentmaya@acmeall models$47.20 of $60 · dayserving
researchdev-teamall models$1,212 of $2,000 · monthserving
cursor-teamdev-teamfable-5, gpt-5.6$980 of $1,500 · monthserving
triageops@acmehaiku-4.5, local$495 of $600 · monthserving
intern-keykai@acmelocal only$25 of $25 · weekpaused at cap

One endpoint from here.

Point your agents at api.experientiallabs.ai/v1 and everything else follows.

Book a call