
- Monthly visits
- 2,268,774
- 2026-09
- Monthly growth +40,559%
Experiential Labs - A unified AI gateway with predictable plan pricing
Experiential Labs provides an OpenAI-compatible gateway to hosted providers, customer provider keys, and local models. Use one API key to route requests, set budgets and access controls, attribute usage, and fail over when a provider is rate-limited or unavailable. Plans include core models within specified request and token limits, with monthly prices.
- LLM Developer Tools
- LLM APIs
- LLM Evaluation
- LLM Fine-Tuning
- AI model gateway
- OpenAI-compatible API
- AI provider routing






Capabilities
Unified model API
Connects compatible clients to models through one OpenAI-compatible endpoint and API key. Point an OpenAI SDK at Experiential, select a model, and send requests; the gateway also serves the OpenAI Responses API and Anthropic Messages API.
Model routing and failover
Routes model requests to providers and can try the next configured provider after a rate limit or outage. Set provider order and allowlists; routing policy can also specify zero-data-retention or no-training providers.
Spend controls and attribution
Sets request budgets and records usage by key, agent, person, team, model, and day. Budgets can stop requests at a configured cap, while request logs show route and cost; customer tags can identify account-level traffic.
Traffic replay and evaluation
Replays supplied traffic against models and scores responses before a model change. Compare model scores and costs using a judge; the homepage describes replaying product traffic as an evaluation workflow.
Traffic-based fine-tuning
Uses enabled prompt and response traces to simulate tasks and fine-tune an open small language model. Training remains off until enabled, the customer starts each run, and resulting weights can be downloaded and self-hosted.
Pricing
Paid plans are recurring monthly subscriptions, and Free has no subscription fee. Compare request and token caps, daily usage, included credits, model discounts, and provider-key options to choose a plan.
| Plan | Price | Included features and limits |
|---|---|---|
| Free | $0 per month |
|
| Pro | $20 per month |
|
| Max | $200 per month |
|
| Ultra | $2,000 per month |
|
Annual payment is available for 12 months paid up front. A temporary $1 verification charge applies to Free and is refunded. Credits are used for requests beyond plan limits at provider list prices.
Use cases
Connect models to an AI product
When an application needs to call different hosted or local models, point an OpenAI-compatible client to the gateway and use one key and endpoint. Choose a model name for each request and access supported providers through the shared interface.
Track and cap model usage
When a product needs visibility into model traffic and spend, assign keys, budgets, and attribution to requests. Inspect logs by route and cost, and apply caps or customer tags to understand and control usage.
Keep model requests available during provider issues
When a selected provider is rate-limited or unavailable, configure an ordered provider route so requests can move to the next provider. The gateway attempts the fallback for the same model, subject to provider availability.
Evaluate or train a model on product traffic
When a team has request traces and wants to compare or adapt models, replay traffic for scored comparisons or enable trace collection for training. Training simulations use the traces, runs are started by the customer, and resulting weights are downloadable.
Limitations
Free core-model limits are shared: 10 requests per minute, 250K tokens per minute, and 1M tokens per day; new Free accounts have separate limits for other models.
Free has no monthly credits; a temporary $1 card-verification charge is refunded and is not added to the credit balance.
Requests using Experiential provider keys are subject to the plan's rate caps; requests using a customer's own provider keys are exempt from Experiential rate limits and billed by that provider.
Requests beyond plan limits can continue only by using credits at the model provider's list price; model pricing and plan discounts vary by model.
Traffic-based training is off until enabled, and each training run must be started by the customer.
Who it is for
Developers integrating AI models into a software product
Teams managing model-provider access, reliability, and usage budgets
Teams building agents that need a shared model endpoint and per-agent controls
Product teams evaluating models or fine-tuning models from their own traces
FAQ
What does unlimited mean on a plan?
Core models have no per-token charge within the plan's limits. Free core models share 10 requests per minute, 250K tokens per minute, and 1M tokens per day, resetting at 00:00 UTC.
Is Free really free?
Free has no subscription fee and no monthly credits. A temporary $1 card-verification charge unlocks free and promotional models; it is refunded and is not added to your credit balance.
Are there rate limits?
Yes. Traffic using Experiential provider keys is capped by plan request and token limits. Free core models share 10 requests and 250K tokens per minute, plus 1M tokens per day; requests above a cap receive a 429 response.
Can I use my own provider keys?
Yes, on Pro and higher. Requests using your own provider keys are billed by the provider and are not rate-limited by Experiential.
Can I pay yearly?
Yes. Annual payment covers 12 months paid up front.
What happens when I exceed plan limits?
Requests can overflow to credits and continue at the provider's list price. Paid plans include monthly credits, while Free has no general credits.
Comments