@@ -68,25 +68,27 @@ current image-capable Grok refs in the bundled catalog.
6868The bundled plugin maps xAI's current public API surface onto OpenClaw's shared
6969provider and tool contracts where the behavior fits cleanly.
707071-| xAI capability | OpenClaw surface | Status |
72-| -------------------------- | -------------------------------------- | ------------------------------------------------------------------- |
73-| Chat / Responses | `xai/<model>` model provider | Yes |
74-| Server-side web search | `web_search` provider `grok` | Yes |
75-| Server-side X search | `x_search` tool | Yes |
76-| Server-side code execution | `code_execution` tool | Yes |
77-| Images | `image_generate` | Yes |
78-| Videos | `video_generate` | Yes |
79-| Batch text-to-speech | `messages.tts.provider: "xai"` / `tts` | Yes |
80-| Streaming TTS | — | Not exposed; OpenClaw's TTS contract returns complete audio buffers |
81-| Speech-to-text | — | Not exposed yet; needs a transcription provider surface |
82-| Realtime voice | — | Not exposed yet; different session/WebSocket contract |
83-| Files / batches | Generic model API compatibility only | Not a first-class OpenClaw tool |
71+| xAI capability | OpenClaw surface | Status |
72+| -------------------------- | ----------------------------------------- | ------------------------------------------------------------------- |
73+| Chat / Responses | `xai/<model>` model provider | Yes |
74+| Server-side web search | `web_search` provider `grok` | Yes |
75+| Server-side X search | `x_search` tool | Yes |
76+| Server-side code execution | `code_execution` tool | Yes |
77+| Images | `image_generate` | Yes |
78+| Videos | `video_generate` | Yes |
79+| Batch text-to-speech | `messages.tts.provider: "xai"` / `tts` | Yes |
80+| Streaming TTS | — | Not exposed; OpenClaw's TTS contract returns complete audio buffers |
81+| Batch speech-to-text | `tools.media.audio` / media understanding | Yes |
82+| Streaming speech-to-text | — | Not exposed; needs streaming transcription contract mapping |
83+| Realtime voice | — | Not exposed yet; different session/WebSocket contract |
84+| Files / batches | Generic model API compatibility only | Not a first-class OpenClaw tool |
84858586<Note>
86-OpenClaw uses xAI's REST image/video/TTS APIs for media generation and the
87-Responses API for model, search, and code-execution tools. Features that need
88-new OpenClaw contracts, such as streaming STT or Realtime voice sessions, are
89-documented here as upstream capabilities rather than hidden plugin behavior.
87+OpenClaw uses xAI's REST image/video/TTS/STT APIs for media generation,
88+speech, and transcription, and the Responses API for model, search, and
89+code-execution tools. Features that need new OpenClaw contracts, such as
90+streaming STT or Realtime voice sessions, are documented here as upstream
91+capabilities rather than hidden plugin behavior.
9092</Note>
91939294### Fast-mode mappings
@@ -239,6 +241,50 @@ Legacy aliases still normalize to the canonical bundled ids:
239241240242</Accordion>
241243244+<Accordion title="Speech-to-text">
245+The bundled `xai` plugin registers batch speech-to-text through OpenClaw's
246+media-understanding transcription surface.
247+248+- Default model: `grok-stt`
249+- Endpoint: xAI REST `/v1/stt`
250+- Input path: multipart audio file upload
251+- Supported by OpenClaw wherever inbound audio transcription uses
252+ `tools.media.audio`, including Discord voice-channel segments and
253+ channel audio attachments
254+255+To force xAI for inbound audio transcription:
256+257+```json5
258+{
259+ tools: {
260+ media: {
261+ audio: {
262+ models: [
263+ {
264+ type: "provider",
265+ provider: "xai",
266+ model: "grok-stt",
267+ },
268+ ],
269+ },
270+ },
271+ },
272+}
273+```
274+275+Language can be supplied through the shared audio media config or per-call
276+transcription request. Prompt hints are accepted by the shared OpenClaw
277+surface, but the xAI REST STT integration only forwards file, model, and
278+language because those map cleanly to the current public xAI endpoint.
279+280+<Note>
281+xAI also offers streaming STT over `wss://api.x.ai/v1/stt`. OpenClaw's
282+bundled xAI plugin does not expose that yet; the current provider is batch
283+STT for file/segment transcription.
284+</Note>
285+286+</Accordion>
287+242288<Accordion title="x_search configuration">
243289The bundled xAI plugin exposes `x_search` as an OpenClaw tool for searching
244290X (formerly Twitter) content via Grok.
@@ -316,9 +362,9 @@ Legacy aliases still normalize to the canonical bundled ids:
316362- `grok-4.20-multi-agent-experimental-beta-0304` is not supported on the
317363 normal xAI provider path because it requires a different upstream API
318364 surface than the standard OpenClaw xAI transport.
319-- xAI STT and Realtime voice are not registered as OpenClaw providers yet.
320- They require transcription/session contracts rather than the existing
321- batch TTS provider shape.
365+- xAI streaming STT and Realtime voice are not registered as OpenClaw
366+ providers yet. Batch xAI STT is registered through media understanding.
367+ Streaming STT and Realtime voice need WebSocket/session contract mapping.
322368- xAI image `quality`, image `mask`, and extra native-only aspect ratios are
323369 not exposed until the shared `image_generate` tool has corresponding
324370 cross-provider controls.
@@ -355,9 +401,10 @@ OPENCLAW_LIVE_TEST=1 OPENCLAW_LIVE_TEST_QUIET=1 OPENCLAW_LIVE_IMAGE_GENERATION_P
355401```
356402357403The provider-specific live file synthesizes normal TTS, telephony-friendly PCM
358-TTS, text-to-image generation, and reference-image editing. The shared image
359-live file verifies the same xAI provider through OpenClaw's runtime selection,
360-fallback, normalization, and media attachment path.
404+TTS, transcribes audio through xAI STT, generates text-to-image output, and
405+edits a reference image. The shared image live file verifies the same xAI
406+provider through OpenClaw's runtime selection, fallback, normalization, and
407+media attachment path.
361408362409## Related
363410此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。