ScriptEngine API Verification & Pricing Methodology

Why this page exists

Clear evidence is more useful than a big promise

ScriptEngine publishes a hosted API and a live model catalog. Catalog entries, upstream providers, package balances, currency rates, and client integrations can change. This page gives developers, buyers, and reviewers a shared way to understand what is current, what is a dated estimate, and what still needs a small test.

It also prevents a common mistake: treating “OpenAI-compatible” as proof that every OpenAI feature is supported. Compatibility is scoped to the endpoints and behaviors that have been tested and documented.

Verified public API surface

What a basic ScriptEngine check covers

The baseline smoke test uses the public base URL https://scriptengine.org/v1, a private bearer key, authenticated model discovery, and one harmless text request. A successful check confirms that the account can reach the service, discover a current model ID, and submit a chat-completions request.

Set the client base URL to https://scriptengine.org/v1.
Send GET /v1/models with Authorization: Bearer.
Select a model returned for the authenticated account; never rely on a copied catalog ID.
Send POST /v1/chat/completions with a short text prompt.
Record the HTTP status, model ID, usage fields, and response time for your own workload.

Tools, embeddings, images, audio, reasoning controls, agent loops, streaming details, retries, fallbacks, and cache behavior are not inferred from this baseline. Test each feature against a current model and client before depending on it.

Pricing method

How token-credit comparisons are calculated

For a model with published input and output credit rates, the comparison keeps input and output separate. We take the package price divided by the included credits to find the cost of one credit, then multiply that unit cost by the model’s input or output credits per million tokens. A workload estimate is then:

input cost  = input tokens  / 1,000,000 × input rate  × credit price
output cost = output tokens / 1,000,000 × output rate × credit price
total       = input cost + output cost

Comparisons use the same model route, token mix, package assumptions, currency conversion date, and fee treatment on both sides. The site’s “up to 86% estimated savings” reference is a dated example, not a universal discount. Actual results vary by model, route, input/output ratio, package, taxes, payment fees, currency, promotions, and upstream changes.

Cached input, batch pricing, reasoning surcharges, image/audio units, minimum charges, and negotiated enterprise rates are excluded unless the source explicitly publishes them and the ScriptEngine route has been independently verified. This keeps a simple token comparison from pretending to be a complete invoice.

Limits and service promises

What ScriptEngine does not currently promise

There is no published uptime or response-time SLA.
Model availability and IDs can change; authenticated discovery is the source of truth.
OpenAI-compatible chat access does not automatically establish support for tools, embeddings, images, audio, Responses API features, retries, fallbacks, or caching.
Price comparisons are estimates tied to a dated snapshot and should be rerun before a purchase or production migration.

These boundaries are useful customer information. They help teams choose ScriptEngine for a verified prepaid API path without confusing it with a full gateway, observability platform, or reliability contract.

Freshness and change control

Dates tell you how much confidence to place in a page

Public pages show a publication or review date when a claim depends on a source snapshot. The model catalog is refreshed separately and should be confirmed with authenticated discovery. When a material page changes, its canonical URL, sitemap lastmod, and llms.txt entry are kept aligned where practical. A date is not a promise that an upstream provider will remain unchanged; it is a signal to re-check.

For competitor comparisons, official pricing or product documentation is linked directly and ScriptEngine’s affiliation is disclosed. “Not publicly documented” means only that the linked source did not establish the claim; it does not assert that a competitor lacks the feature.

Reproduce a check

Use your own key and workload

Do not paste credentials into a browser address bar, support ticket, prompt, or public repository. The following examples use an environment variable and intentionally omit a secret and model ID:

curl https://scriptengine.org/v1/models \
  -H "Authorization: Bearer $SCRIPTENGINE_API_KEY"

curl https://scriptengine.org/v1/chat/completions \
  -H "Authorization: Bearer $SCRIPTENGINE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"MODEL_ID_FROM_DISCOVERY","messages":[{"role":"user","content":"Reply with one word: test"}]}'

For purchasing decisions, repeat the test with the model, prompt size, output length, concurrency, and error-handling pattern that matter to you. See the plans, capability matrix, support page, and operator information alongside this methodology.

ScriptEngine API verification and pricing methodology