At a glance
Current capability status
“Verified” means the behavior is part of the current ScriptEngine integration contract and has a documented smoke-test path. “Model-specific” means the client or upstream protocol may support it, but the feature must be tested for the exact model and request shape. “Not published” means ScriptEngine does not currently provide a public claim that should be used as a purchasing assumption.
| Capability | Status | What to do |
|---|---|---|
| Bearer API authentication | Verified | Use a private key in an environment variable or secret manager. |
| Authenticated model discovery | Verified | Call GET /v1/models and choose a returned ID. |
| Chat completions | Verified | Use POST /v1/chat/completions with the discovered model ID. |
| OpenAI-compatible base URL | Verified | Use exactly https://scriptengine.org/v1. |
| Streaming | Model-specific | Test the exact model and client before relying on streamed output or usage data. |
| Tool/function calling | Not globally verified | Do not enable tools solely because a client exposes the option. |
| Embeddings | Not publicly verified | Use a separately tested embedding provider unless ScriptEngine publishes a confirmed route. |
| Images, audio, or video | Not publicly verified | Do not infer multimodal support from the model catalog name. |
| Responses or Open Responses API | Not globally verified | Keep integrations on chat completions until a route is tested and documented. |
| Prompt-cache pricing | Not publicly published | Compare standard input/output rates only unless current cache fields are supplied. |
| Retries, routing, or fallback | Not publicly promised | Implement application-level timeout and fallback behavior where needed. |
| Uptime or response-time SLA | None published | Treat this as a tested API path, not a contractual reliability guarantee. |
Reproducible check
Run the smallest useful API test
Use the same key for discovery and chat. This confirms account authentication, model availability, and the request path without exposing a secret in a browser or repository.
export SCRIPTENGINE_API_KEY="YOUR_PRIVATE_KEY"
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 exactly: ScriptEngine connected"}]}'A successful text response does not certify tools, images, embeddings, caching, or SLA behavior. Record the date, model ID, request type, and observed result when deciding whether to expand the integration.
Client paths
Choose a documented setup workflow
Each guide below is based on the client’s current official configuration documentation and keeps the ScriptEngine claim scoped to the verified surface.
OpenClaw
Custom provider JSON5, model discovery, and a text-first smoke test.
Open guide → Chat UIOpen WebUI
OpenAI-compatible connection setup and model refresh.
Open guide → CodingCline
OpenAI Compatible provider setup before repository permissions.
Open guide → CodingContinue
YAML model configuration with chat role only at first.
Open guide → PythonLangChain
ChatOpenAI base URL and a plain Python invocation.
Open guide →Cost and availability
Use the live catalog for rate decisions
ScriptEngine’s public model catalog and prepaid plans are rate references, not a promise that every model is available to every key forever. Model IDs and route details can change. The published savings methodology shows a dated reference maximum of up to 86% for one comparison; it is not a universal discount across all models, inputs, outputs, cache states, or currencies.
Before production, discover the models with the key you will actually use, run a representative request, and record the selected route. This is also the fairest way to compare ScriptEngine with an affordable OpenRouter alternative or another gateway.
Questions developers ask
Capability answers without guesswork
Is ScriptEngine fully compatible with every OpenAI API feature?
No. The public, verified surface is bearer authentication, model discovery, and chat completions. Other protocol features require separate evidence.
Can I use this page as a promise of tool calling?
No. A client’s tool or agent option is not proof that the selected ScriptEngine model accepts the same request shape.
Where should I report an integration problem?
Use the ScriptEngine support route and include the endpoint, status code, model ID, and a redacted request description. Never send the full API key.
Next step
Connect from the documented surface
Read the API integration overview, choose a model from authenticated discovery, and use the matching client guide. Re-check this matrix after major API or model-catalog changes.