Glossary
Common concepts, fields, and error terms used in the help docs
This document explains common concepts and fields used throughout the help docs.
Access & authentication
| Term | Description |
|---|---|
| API Key / Token | Access key starting with sk-, passed via Authorization: Bearer |
| Base URL | Gateway deployment address, e.g. https://www.guid.ai |
| Bearer Token | HTTP auth scheme: Authorization: Bearer sk-xxx |
| Session | Admin console login session, used for /api/* and some content download endpoints |
Models
| Term | Description |
|---|---|
| model | Model ID in the request body or path; must be a data[].id returned by /v1/models |
| supported_endpoint_types | Endpoint types supported by the model; determines which capabilities it can use |
openai | Supports Chat / Completions |
openai-response | Supports /v1/responses |
image-generation | Supports /v1/images/* |
openai-video | Supports video generation endpoints |
| owned_by | Model ownership identifier (provider ID) |
| tags | Model tags (if any) |
Text
| Term | Description |
|---|---|
| SSE | Server-Sent Events streaming format, Content-Type: text/event-stream |
| stream | Request field; when true, returns an SSE stream |
| finish_reason | Why generation stopped: stop, length, tool_calls, content_filter |
| tool_calls | Function/tool calls initiated by the model |
| usage | Token usage object, including prompt_tokens, completion_tokens, etc. |
Images
| Term | Description |
|---|---|
| response_format | Image response format: url (default) or b64_json |
| revised_prompt | Prompt rewritten by some models |
| quota | Per-request billing consumption (image usage.quota) |
Videos
| Term | Description |
|---|---|
| task_id | Public video task ID, shaped like task_xxxxxxxx |
| Async task | Returns a task_id immediately on create; poll status, then download |
| status | Task status: queued → in_progress → completed / failed |
| progress | Task progress, 0–100 |
| mode | Generation mode for guidai multimodal video (t2v, first_frame, ref_image, etc.) |
| assets | Asset array in guidai requests, [{ "role": "...", "url": "..." }] |
| variant | Content download parameter: video (default) or thumbnail |
| remixed_from_video_id | Video result URL returned by some channels when a task completes |
guidai video modes
| mode | Description |
|---|---|
t2v | Text-to-video |
i2v | Image-to-video |
first_frame | Generate video from a first frame |
first_last_frame | Generate video from first and last frames |
ref_image | Generate video from reference images |
video_extend | Extend / continue a video |
video_edit | Smart video editing |
Audio
| Term | Description |
|---|---|
| voice | TTS voice code from /v1/audio/voice/{model} → data[].code |
| response_format | Audio output format: mp3, wav, opus, etc. |
| Channel voices | name equals code; comes from channel configuration |
| Platform voice library | name is the display name; code is voice_key |
Skills
| Term | Description |
|---|---|
| Skills | Skill package based on the V1 API (guid-ai-skills), includes SKILL.md, for in-conversation Agent calls |
| guid-ai-skills | Official Guid AI skill package name covering image, video, and audio generation |
| GUIDAI_BASE_URL | Environment variable for the Skills gateway URL |
| GUIDAI_API_KEY | Environment variable for the Skills API Key |
| Supported platforms | Agent clients such as Claude, Cursor, Codex, OpenClaw, Hermes, and NemoClaw |
MCP
| Term | Description |
|---|---|
| MCP | Model Context Protocol; another Agent integration path that calls Tools via the /mcp endpoint |
| Tool | Callable capability exposed by MCP, e.g. generate_image, create_video |
| mcpServers | Client MCP config key; recommended name is guid |
| Mcp-Session-Id | Streamable HTTP session ID; must be included on subsequent requests |
| next_poll_after_sec | Suggested video polling interval (default 10 seconds) |
Errors
| Term | Description |
|---|---|
| Standard error | { "error": { "message", "type", "code" } } format |
| Task error | { "code", "message" } format, common for video tasks |
| Gateway extended error | { "success": false, "message" } format, common for model lists |
See Common Conventions · Error Responses.
Billing
| Term | Description |
|---|---|
| Token billing | Billed by input/output tokens; seen in text usage |
| Per-request billing | Billed by request count; seen in image usage.quota |
| Per-second billing | Video may be billed by generated duration (depends on model configuration) |