Pinned Tweet
I grew up with a borderline unhealthy obsession with SpaceX. I didn’t miss a single Falcon 9 launch in high school, sometimes waking up at 4 AM to watch the booster touchdown. I remember when the original Tintin sats launched, Bob and Doug’s flight on Demo-2, and the very first Transporter mission. I religiously followed every update from the Starhopper test campaign at Boca.
Today, SpaceX has proven to the world that a fully reusable rocket is possible.
Every day, I’m grateful to be able to contribute to this mission in a small and indirect way. What other company makes you feel this way?
13
301
Orbital patina.
Ship 42 on final descent into the chopsticks at Pad 2 under power from its sea-level engines after returning from orbit.
Having launched yesteday with Booster 22, Ship 42 spent a total of just over 22 hours in space.
📸 - @NASASpaceflight
things are only gonna move faster from here 🚀
damn voice AI had a crazyyy month:
1. @grok Grok Voice 3.0: hit 97% on Full Duplex Bench. also being used in Tesla and Starlink.
2. @phonic_co Phonic-1: can call tools while you're still talking. sub-600ms response times and was a top contender on Artificial Analysis.
3. @blocks Dessa: trained entirely on human conversation audio. first demo where I genuinely couldn't tell who was the human and who was AI. though...I'm curious how smart it is.
4. @FishAudio Fish Audio Live 3: trained on real conversations to pick up on natural turn-taking, like when to take over the turn, backchannel, or let someone finish speaking. can react in as little as 186ms.
seems like there's a lot of focus on how we can make AI converse like a human, not just have the voice itself sound natural
4
346
Larry Jing retweeted
Some small improvements to the voice experience in Grok @bot!
You can now changes your bots voice and speed via chat. This can be used for outbound calling as well. Simply ask your bot!
@johnbai and I revamped the voice settings and gave more speed options.
We are rolling out improvements to general bot voice intelligence and making it interrupt less.
Please let me know any other requests/feedback for voice 🫶
One of the coolest products ever
Introducing Mini 2, the next-generation portable Starlink. Get up to 4.5 hours of connectivity on a single charge with the quick-swap battery, or plug in with USB-C.
With built-in Wi-Fi, all you need is a clear view of the sky to get online in 60 seconds.
Preorders are now open by invitation in the US. Be the first to know when it's available → starlink.com/mini-2
Grok Voice 3.0 is now live on Retell! 🚀
@SpaceXSI's speech-to-speech model combines sub-second latency with natural, concise responses, ranking #1 in colloquial performance out of 41 models on our vCX Benchmark.
Every voice AI use case is different. That’s why Retell gives teams the freedom to choose the right models for their AI voice agents, backed by real performance data.
Explore the full benchmark: bench.retellai.com/
Try asking Grok @Bot to make calls!
Grok @Bot's new calling feature is very impressive.
You just tell it who to call and why, and it handles the conversation for you. It shows a real-time transcript of the call, and if Grok needs additional info from you, a little answer box pops up where you type/select your response, and Grok relays it to the person on the phone. Pretty incredible.
Here's an example (with the call audio) of what's it's like to use:
Larry Jing retweeted
We are rolling out a new streaming dictation experience in Grok @bot! Available in mobile on 1.18.
Built upon the exceptional work of @aegeantic. Video made by my bot!
Roadster xitter.gq/i/broadcasts/1yKAPwNgg…
0:00
Tesla
Roadster
Larry Jing retweeted
BREAKING: Grok Voice 3.0 by @SpaceXSI lands at #1 on AudioAgentBench, Design Arena's speech-to-speech leaderboard, with a score of 85.1%.
This is an impressive 7.4-point increase from Grok Voice Think Fast 2.0, the previous #1. This also places xAI newest model among leading voice models such as GPT Live 1 by @OpenAI.
Congratulations to the @SpaceXSI team on taking the top spot!
Grok Voice 3.0 is now the world’s leading conversational voice model
SpaceXAI’s Grok Voice 3.0 is the new leading Speech to Speech model, debuting at #1 on the Artificial Analysis Speech to Speech Index at 86.9, while also being competitive on latency
Grok Voice 3.0 is @SpaceXSI 's successor to Grok Voice 2.0, a Speech to Speech model available through the SpaceXAI Realtime API at $0.08 per minute. We evaluated Grok Voice 3.0 with the Rex voice through the SpaceXAI Realtime API.
Key takeaways:
➤ Speech to Speech Index: Grok Voice 3.0 debuts at #1 at 86.9, ahead of GPT-Live-1 (Astra, medium) at 83.2, Gemini 3.8 Live Extended Thinking (High) at 82.6 and GPT-Live-1 (Sol, low) at 80.1. It is up on Grok Voice 2.0 at 81.3 (+5.6 points) and Grok Voice 1.0 at 73.7 (+13.2 points), released just under six months ago. The Index averages Speech Reasoning (Big Bench Audio), Agentic Performance (Tau Voice), Arena Preference, and Task Success Rate
➤ Speech Agent Arena: Grok Voice 3.0 ranks #3 in preference at Elo 1070, behind Gemini 3.1 Flash Live Minimal (1100) and Gemini 3.8 Live (1088) and ahead of GPT-Live-1 (Sol, low) at 1057, and #1 on Task Success Rate at 94.8%, highest on the board, just ahead of Grok Voice 2.0 at 94.6%. Grok Voice 2.0 trails at Elo 1011
➤ Tau Voice: Grok Voice 3.0 takes the #2 spot on our Tau Voice benchmark implementation at 71.6%, behind GPT-Live-1 (Astra, medium) at 74.5% and ahead of Gemini 3.8 Live Extended Thinking (High) at 68.6% and GPT-Live-1 (Sol, low) at 59.3% - up from 56.5% for Grok Voice 2.0
➤ Big Bench Audio: Grok Voice 3.0 scores 98.5% on audio reasoning, ahead of Gemini 3.8 Live Extended Thinking (High) at 97.7% and Grok Voice 2.0 at 97.2%
➤ Speed: Average Time to First Audio on Big Bench Audio is 0.81 seconds for Grok Voice 3.0, well ahead of GPT-Live-1 (Sol, low) at 1.24s, GPT-Live-1 (Astra, medium) at 1.34s and Gemini 3.8 Live Extended Thinking (High) at 1.35s, though slower than Grok Voice 2.0 at 0.70s
➤ Cost: Grok Voice 3.0 costs $4.80 per hour of input audio, the same as Grok Voice 2.0. That is above GPT-Live-1 (Sol, low) at $4.47 and Gemini 3.8 Live Extended Thinking (High) at $3.50, cheaper than GPT-Live-1 (Astra, medium) at $5.83, and ~2.2x cheaper than GPT-Realtime-2.1 High at $10.75
Congrats to @elonmusk and the @SpaceXSI team 🚀
See below for more detail ⬇️
Larry Jing retweeted
SpaceXAI’s Grok Voice 3.0 is the new leading Speech to Speech model, debuting at #1 on the Artificial Analysis Speech to Speech Index at 86.9, while also being competitive on latency
Grok Voice 3.0 is @SpaceXSI 's successor to Grok Voice 2.0, a Speech to Speech model available through the SpaceXAI Realtime API at $0.08 per minute. We evaluated Grok Voice 3.0 with the Rex voice through the SpaceXAI Realtime API.
Key takeaways:
➤ Speech to Speech Index: Grok Voice 3.0 debuts at #1 at 86.9, ahead of GPT-Live-1 (Astra, medium) at 83.2, Gemini 3.8 Live Extended Thinking (High) at 82.6 and GPT-Live-1 (Sol, low) at 80.1. It is up on Grok Voice 2.0 at 81.3 (+5.6 points) and Grok Voice 1.0 at 73.7 (+13.2 points), released just under six months ago. The Index averages Speech Reasoning (Big Bench Audio), Agentic Performance (Tau Voice), Arena Preference, and Task Success Rate
➤ Speech Agent Arena: Grok Voice 3.0 ranks #3 in preference at Elo 1070, behind Gemini 3.1 Flash Live Minimal (1100) and Gemini 3.8 Live (1088) and ahead of GPT-Live-1 (Sol, low) at 1057, and #1 on Task Success Rate at 94.8%, highest on the board, just ahead of Grok Voice 2.0 at 94.6%. Grok Voice 2.0 trails at Elo 1011
➤ Tau Voice: Grok Voice 3.0 takes the #2 spot on our Tau Voice benchmark implementation at 71.6%, behind GPT-Live-1 (Astra, medium) at 74.5% and ahead of Gemini 3.8 Live Extended Thinking (High) at 68.6% and GPT-Live-1 (Sol, low) at 59.3% - up from 56.5% for Grok Voice 2.0
➤ Big Bench Audio: Grok Voice 3.0 scores 98.5% on audio reasoning, ahead of Gemini 3.8 Live Extended Thinking (High) at 97.7% and Grok Voice 2.0 at 97.2%
➤ Speed: Average Time to First Audio on Big Bench Audio is 0.81 seconds for Grok Voice 3.0, well ahead of GPT-Live-1 (Sol, low) at 1.24s, GPT-Live-1 (Astra, medium) at 1.34s and Gemini 3.8 Live Extended Thinking (High) at 1.35s, though slower than Grok Voice 2.0 at 0.70s
➤ Cost: Grok Voice 3.0 costs $4.80 per hour of input audio, the same as Grok Voice 2.0. That is above GPT-Live-1 (Sol, low) at $4.47 and Gemini 3.8 Live Extended Thinking (High) at $3.50, cheaper than GPT-Live-1 (Astra, medium) at $5.83, and ~2.2x cheaper than GPT-Realtime-2.1 High at $10.75
Congrats to @elonmusk and the @SpaceXSI team 🚀
See below for more detail ⬇️
Intelligence and latency often come at a tradeoff for voice models - the graph below shows where the rest of the industry sits.
Grok Voice 3.0 refuses to pick between the two. Say hello to the new Pareto frontier!
Replying to @SpaceXSI
Grok Voice 3.0 achieves frontier performance in voice agent tasks at 40% lower latency. At SpaceX, it resolves 70% of Starlink support calls on its own.























