Skip to content
  • Tech Dev NotesTech Dev Notes
Apps
  • Lookup
  • Compare
Market movement
  • Charts
  • Rankings
Visual proof
  • Screens
  • Store listings
  • Icons
Stacks and releases
  • Tech stacks
  • Tool releases
  • Developers
More
  • X feature flags
  • Grokipedia
  • Grok companions
  • Blog
  • Follow on X
ChartsRankingsMoversCompareLookupScreensTech stacksReleases

© 2026 Tech Dev Notes

RSSAboutAPIStatusSecurityPrivacyTermsSitemap@techdevnotes
xAI Docs

xAI · Docs

xAI Docs

August 25, 2026 at 11:17 PM UTC

Added an FAQ covering SIP call control:

Changelog

Audio

Speech to Speech — SIP (/developers/model-capabilities/audio/speech-to-speech/sip)

Added an FAQ covering SIP call control:

  • How POST /v1/realtime/calls/{call_id}/refer reports transfer success/failure (including SIP reason codes)
  • Session behavior when a transfer fails
  • Reattaching a resumed conversation to a new SIP call via session resumption
  • Transferring to a Twilio SIP Domain

Speech to Text (/developers/model-capabilities/audio/speech-to-text)

  • Documented Opus streaming (encoding=opus): one raw Opus packet per binary frame, supported sample rates, mono-only, and bandwidth guidance
  • Streaming query params updated for opus; default endpointing documented as 400 ms
  • Clarified that most errors (including undecodable frames) close the connection

REST API — Speech to Text & Voice

Same streaming updates on:

  • /developers/rest-api-reference/inference/speech-to-text
  • /developers/rest-api-reference/inference/voice

Images & video

Image generation (/developers/model-capabilities/images/generation)

  • Clarified n (1–10) across REST/OpenAI SDKs vs xAI Python sample() / sample_batch()
  • Documented aspect_ratio default auto, expanded ratio table, and resolution (1k default, 2k)
  • Added quality examples and clarified response_format (url default vs base64)

Video generation (/developers/model-capabilities/video/generation)

  • Documented generate_audio=false for silent videos (audio track remains the default)

Reference to video (/developers/model-capabilities/video/reference-to-video)

  • Expanded multi-image / multi-voice examples (reference_images + reference_audios with <IMAGE_*> / <AUDIO_*> prompt tags)

Batch API

Batch API (/developers/advanced-api-usage/batch-api)

  • Image/video batch examples now use grok-imagine-image-2.0 and grok-imagine-video-1.5
  • Batch-enabled model guidance points to the models pages instead of a fixed allowlist

Grok Bot

Bots (/grok-bot/bots)

New Share a Bot section: public share links, preview on x.ai, copy-on-add behavior, and what the link exposes

FAQ (/grok-bot/faq)

Added “Can I share a Bot with someone else?” with link to sharing docs and third-party bot terms

Approvals, security, and privacy (/grok-bot/approvals-security-and-privacy)

Noted that sharing is not a security boundary—strip secrets before sharing

Older release← Aug 25, 19:17 UTCNewer release→ Aug 26, 00:17 UTC
All xAI Docs releases →