Voice agents now on Vercel
Voice agents, now on Vercel.
Realtime, speech and transcription are now live on AI Gateway. Build with 𝚞𝚜𝚎𝚁𝚎𝚊𝚕𝚝𝚒𝚖𝚎, 𝚐𝚎𝚗𝚎𝚛𝚊𝚝𝚎𝚂𝚙𝚎𝚎𝚌𝚑 & 𝚝𝚛𝚊𝚗𝚜𝚌𝚛𝚒𝚋𝚎 on AI SDK 7.
Vercel's AI SDK 7 provides the underlying useRealtime hook
Voice agents now on Vercel
Voice agents, now on Vercel.
Realtime, speech and transcription are now live on AI Gateway. Build with 𝚞𝚜𝚎𝚁𝚎𝚊𝚕𝚝𝚒𝚖𝚎, 𝚐𝚎𝚗𝚎𝚛𝚊𝚝𝚎𝚂𝚙𝚎𝚎𝚌𝚑 & 𝚝𝚛𝚊𝚗𝚜𝚌𝚛𝚒𝚋𝚎 on AI SDK 7.
Positive users praise Vercel's integration of Grok realtime voice models into its AI Gateway for enabling fast and easy voice agent builds, while negative users point to added latency that breaks immersion and reduces retention.
No Digg Deeper questions have been answered for this story yet.
Try SpaceXAI Voice models in the Vercel AI Gateway
Grok's realtime voice is now on AI Gateway. Build with AI SDK 7:
• 𝚡𝚊𝚒/𝚐𝚛𝚘𝚔-𝚟𝚘𝚒𝚌𝚎-𝚝𝚑𝚒𝚗𝚔-𝚏𝚊𝚜𝚝-𝟷.𝟶 (𝚞𝚜𝚎𝚁𝚎𝚊𝚕𝚝𝚒𝚖𝚎) • 𝚡𝚊𝚒/𝚐𝚛𝚘𝚔-𝚝𝚝𝚜 (𝚐𝚎𝚗𝚎𝚛𝚊𝚝𝚎𝚂𝚙𝚎𝚎𝚌𝚑) • 𝚡𝚊𝚒/𝚐𝚛𝚘𝚔-𝚜𝚝𝚝 (𝚝𝚛𝚊𝚗𝚜𝚌𝚛𝚒𝚋𝚎)
https://vercel.com/changelog/xai-grok-audio-models-now-available-on-vercel-ai-gateway

@xai

@parker__conrad Great voice. We tested all and the team loved Grok’s
State of the art voice APIs from SpaceXAI, now in the Vercel AI Gateway
Grok's realtime voice is now on AI Gateway. Build with AI SDK 7:
• 𝚡𝚊𝚒/𝚐𝚛𝚘𝚔-𝚟𝚘𝚒𝚌𝚎-𝚝𝚑𝚒𝚗𝚔-𝚏𝚊𝚜𝚝-𝟷.𝟶 (𝚞𝚜𝚎𝚁𝚎𝚊𝚕𝚝𝚒𝚖𝚎) • 𝚡𝚊𝚒/𝚐𝚛𝚘𝚔-𝚝𝚝𝚜 (𝚐𝚎𝚗𝚎𝚛𝚊𝚝𝚎𝚂𝚙𝚎𝚎𝚌𝚑) • 𝚡𝚊𝚒/𝚐𝚛𝚘𝚔-𝚜𝚝𝚝 (𝚝𝚛𝚊𝚗𝚜𝚌𝚛𝚒𝚋𝚎)
https://vercel.com/changelog/xai-grok-audio-models-now-available-on-vercel-ai-gateway

@xAI @grok @Support @elonmusk
My journey with Grok started in January 2026.
At first, there were speed and performance issues. Then xAI introduced message limits. I accepted that because Grok actually became more stable.
Later, hourly limits were added: "Your limit will reset in 2 hours." Again, I understood the reason.
Now, weekly limits have also been introduced.
However, the Grok website says that once your limit is reached, you can either purchase credits or continue using the free features. It also states that users should receive 10–20 free messages every two hours.
It's now been over 16 hours since I hit my weekly limit, and those free messages have never appeared.
I'm a **SuperGrok** subscriber. I've contacted support by email ([support@x.ai](mailto:support@x.ai)) and through the app, but after 16 hours I still haven't received any response.
I truly appreciate the work of the xAI team. Building and maintaining an AI system like Grok is incredibly difficult, and I respect that.
But if new limits are introduced, they should work exactly as described. Paid subscribers deserve clear communication, reliable service, and support when something goes wrong.
One more thing surprised me.
When I wanted to post this message as a reply to Elon Musk and other people involved with Grok, X told me that I would need an X Premium subscription just to publish a longer comment.
That made me wonder: doesn't this discourage open discussion? Everyone can post a short opinion, but if someone wants to explain an issue in detail, they have to pay for the ability to do so.
Has anyone else experienced the same problems with Grok's limits or support?
#Grok #xAI #SuperGrok #AI #ArtificialIntelligence #CustomerExperience #Feedback #Tech #ElonMusk #X

@rauchg Right on time for OpenDex – a voice-first open source agentic harness.
Was working on a Login with Vercel for it :)
https://github.com/wassgha/opendex

@milichab @elonmusk It’s To Late!

@milichab voice latency is the real test here. gateway is nice only if streaming feels instant enough.

@milichab @elonmusk Grok 1.5 ✨
Grok Voice in Vercel!
Grok's realtime voice is now on AI Gateway. Build with AI SDK 7:
• 𝚡𝚊𝚒/𝚐𝚛𝚘𝚔-𝚟𝚘𝚒𝚌𝚎-𝚝𝚑𝚒𝚗𝚔-𝚏𝚊𝚜𝚝-𝟷.𝟶 (𝚞𝚜𝚎𝚁𝚎𝚊𝚕𝚝𝚒𝚖𝚎) • 𝚡𝚊𝚒/𝚐𝚛𝚘𝚔-𝚝𝚝𝚜 (𝚐𝚎𝚗𝚎𝚛𝚊𝚝𝚎𝚂𝚙𝚎𝚎𝚌𝚑) • 𝚡𝚊𝚒/𝚐𝚛𝚘𝚔-𝚜𝚝𝚝 (𝚝𝚛𝚊𝚗𝚜𝚌𝚛𝚒𝚋𝚎)
https://vercel.com/changelog/xai-grok-audio-models-now-available-on-vercel-ai-gateway

@milichab @elonmusk It’s not just about voice models — it’s about how Vercel is quietly becoming the distribution layer for AI inference.

@milichab 현실적인 조언 감사합니다 💯

@milichab This confirms everything! Internet infrastructure stress-tested.

@milichab @elonmusk Just what I needed, another voice in my head to ignore. Can’t wait to see if it’s as chaotic as the real thing.

@milichab @elonmusk 🌎

@xai I swapped Alexa to Eve in my Home Assistant announcements using 𝚡𝚊𝚒/𝚐𝚛𝚘𝚔-𝚝𝚝𝚜, and output to my Sonos speakers. Now need to figure out a way to issue voice commands via xai. Indeed all done with Grok Build Composer without writing a single line of code or yaml.

@milichab https://youtu.be/IH652iPlLa0?si=2fFeBddlOOiYfJHh

@xai time to yap 🗣️

@rauchg @rauchg's voice via @ElevenLabs when? 👀

@milichab Oh we shippin