Choose an API, SDK, or CLI¶
First choose whether you need to call models or manage resources. Then choose an API, SDK, or CLI based on where the code runs and how much control you need.
Quick selection¶
Task |
Recommended option |
Why |
|---|---|---|
Call model chat, Responses, embeddings, or reranking from an application |
Genesis Model API |
Use the model interface and response shape directly. |
Call compatible interfaces with an existing OpenAI or Anthropic client |
Genesis SDK |
Keep a familiar SDK calling pattern while using Genesis credentials and service address. |
Control HTTP methods, headers, fields, and responses precisely |
AI Studio API |
Suitable for service-to-service integrations, webhooks, and interfaces without an SDK wrapper. |
Operate workspace resources from a Go or Python application |
AI Studio SDK |
Provides typed clients, workspace-scoped objects, and task wrappers. |
Operate workspace resources in a terminal, operations script, or CI |
|
Suitable for interactive operations and automation; it does not call Genesis models. |
Integration configuration¶
API¶
API |
Service address |
Credentials and scope |
|---|---|---|
The complete Genesis service address copied from the console Use page |
Bearer credentials; model IDs come from the model list available to the current credential. |
|
|
Pass a personal access token in |
SDK¶
SDK |
Service address |
Credentials and scope |
|---|---|---|
The complete Genesis service address copied from the console Use page |
Bearer credentials; model IDs come from the model list available to the current credential. |
|
Configure the Genesis origin/root as described on the Anthropic SDK page; the SDK appends the Messages path |
Pass the Genesis credential with |
|
The same service address as the AI Studio API |
Personal access token; list workspaces first, then create a workspace-scoped object. |
CLI¶
CLI |
Service address |
Credentials and scope |
|---|---|---|
The same service address as the AI Studio API |
Personal access token; run |
moi-cli does not call Genesis models. AI Studio SDK is not an OpenAI- or Anthropic-compatible SDK; it operates workspace-scoped resources.
If you are still unsure¶
Start with the AI Studio API or Genesis API reference when you need to verify the exact HTTP contract.
Evaluate whether AI Studio SDK already wraps the task when an application must coordinate multiple resource objects.
Use
moi-clifor interactive inspection or for placing a small number of commands in a script.Return to the AI Studio API when the SDK or CLI does not provide the required capability. Do not infer an interface from a command or method name.
Confirm before calling¶
Whether the object already has a public API, SDK method, or CLI command.
Whether the operation modifies a definition object or creates a one-off asynchronous run.
Whether the success response returns an object ID, a run ID, or a request ID that you need to save.
Which status or result endpoint to query afterward, and what result counts as business completion.
Whether delete, cancel, retry, or resubmit can leave external side effects.
Interface capabilities not listed in the developer documentation should not be inferred to be stable public interfaces from web requests, internal paths, or similar commands.
Next steps¶
Call Genesis models:
Manage AI Studio resources: