Pinned Tweet
i left @spacexai today.
the last almost 2 yrs have been incredibly fun building Grok, Starfleet, Grokipedia, X and more. we achieved the impossible in the shortest amount of time.
as a 21yr old in silicon valley there’s always been a whisper in my ear to someday build my own thing. i can’t ignore it anymore.
thank you elon and the whole team for changing my life (and letting me be [email protected]). genuinely love every single person i worked with. more to come.
sic itur ad astra.
i feel like a lot of these ai personal assistants will be crushed by apple once they decide to lock in on siri. not cause they have better talent or capital but cause integration with the apple ecosystem (esp iMessage) is such a huge advantage.
there is still a lot of room in space for specialized/enterprise ai assistants like grokbot or claude code.
Raman Khatri retweeted
Great to see sports coming together on X, seems so obvious now not so much when it was starting out. @AndrewMusk @lucasgordon @aqueel @ramankhatri @shlomiatar teapawt.net/i/nfl/game/20528083792…
We're publishing our most detailed threat intelligence report to date.
It covers how people tried to misuse Claude—for cyberattacks, influence operations, surveillance, biology, and building weapons—and how we found and stopped them.
We disrupted every operation in the report, and used the lessons from them to strengthen our safeguards. Where appropriate, we also shared what we found with authorities and other AI companies.
These cases are not typical: we’re highlighting some of the most sophisticated misuse we’ve seen. But they’re especially important to discuss, because they show us where AI misuse is headed, where our safeguards work, and where they need to improve.
We’re publishing this report so others can spot the same activity on their own platforms, and so we can give the public a clearer view of how emerging threats develop.
Read the report: anthropic.com/threat-intelli…
"Zuckerberg's Wang is penetrating the benchmarks, edging on the frontier"
Meta has released Muse Spark 1.3, their fourth Muse Spark model release in five months. Muse Spark 1.3 (max), which is in limited preview for Meta’s partners, scores 62 on the Artificial Analysis Intelligence Index, behind only Claude Fable 5.1 and Claude Opus 5. The variant available now, Muse Spark 1.3 (xhigh), scores 61 and ties with GPT-5.6 Sol (max) and Grok 4.6 (high). Both variants’ gains come primarily from improvements in agentic work and scientific capabilities
Muse Spark 1.3 (xhigh) enters the Artificial Analysis Intelligence Index at 61, up 4 points from Muse Spark 1.2 (57, August) and 8 points from Muse Spark 1.1 (53, July). It enters tied with GPT-5.6 Sol (max), Grok 4.6 (high), and Claude Opus 5 (high), and behind Claude Fable 5.1 (max, 66), Claude Opus 5 (max, 63), and Claude Fable 5 (max, 62)
Muse Spark 1.3 (max), which is in a limited preview stage, lands at 62. This higher index score is enabled by gains vs. Muse Spark 1.3 (xhigh) in Tau3-Bench Banking (52% vs. 47%) and GDPval-AA v2 (1,754 Elo vs. 1,709). Muse Spark 1.3 (max) is second only to Claude’s Fable and Opus variants in total score
Congratulations to @AIatMeta, @finkd, and @alexandr_wang on the release!
Key Takeaways:
➤ Continued improvement on agentic knowledge work tasks. At the launch of Muse Spark 1.2, we noted its significant gains in agentic knowledge work performance vs. Muse Spark 1.1. The latest iteration continues this trend, with Muse Spark 1.3 (xhigh) demonstrating a notable 12-point gain vs. Muse Spark 1.2 in Tau3-Bench Banking (35% to 47%), a 5-point gain in Terminal-Bench 2.1 (80% to 85%), and a new GDPval-AA v2 Elo of 1709 against its predecessor’s 1615. Muse Spark 1.3 (max) improves further on Tau3-Bench Banking (52%) and GDPval-AA v2 (1,754 Elo). This Tau3-Bench Banking score is #1 among all models. Muse Spark 1.3 (max) achieves these higher agentic work scores by using more turns and total reasoning tokens, reasoning 62% more on GDPval-AA v2 and 28% more on Tau3-Bench Banking compared to Muse Spark 1.3 (xhigh)
➤ The lowest cost per task for any model at 59+ on the Artificial Analysis Intelligence Index. Muse Spark 1.3 (xhigh) costs $0.55 per Intelligence Index task at Meta's unchanged $1.25/$4.25 per 1M token pricing ($0.15 for cached input), with its peers GPT-5.6 Sol (max) and Grok 4.6 (high) costing $0.95 and $0.94 respectively, a 70%+ premium. This places Muse Spark 1.3 (xhigh) on the Pareto frontier for Intelligence vs. Cost per Task. Its cost per task is higher than Muse Spark 1.2 ($0.40 per task), driven by ~57% more input tokens per task on agentic evaluations, with output tokens up only ~8%. Pricing for Muse Spark 1.3 (max) is not yet publicly available
➤ Scientific Reasoning results rose across the board, led by CritPt. CritPt was the standout non-agentic score gain vs. Muse Spark 1.2, with a material +8 points for the xhigh variant (18% to 26%), and GPQA Diamond achieved +4 points (90% to 94%), while Humanity’s Last Exam and SciCode each gained a more modest 2-3 points (45% to 47% and 56% to 59%, respectively). Muse Spark 1.3 (max) achieved roughly similar scores to the xhigh variant, gaining 2 points in Humanity’s Last Exam, tying on GPQA Diamond, and losing a point on CritPt vs. Muse Spark 1.3 (xhigh)
➤ Minor regressions in only two evaluations. Both Muse Spark 1.3 (xhigh) and Muse Spark 1.3 (max) dropped 4 points in AA-LCR (83% to 79%) when compared to Muse Spark 1.2, and AA-Omniscience (Accuracy) fell 3 points for xhigh and 1 point for max. The drops in AA-Omniscience (Accuracy) are due to a higher abstention rate (not answering questions when unsure), which also lowered the hallucination rate for Muse Spark 1.3 (xhigh)
Other model details (xhigh variant):
➤ Context window: 1M tokens, unchanged from Muse Spark 1.2
➤ Pricing: unchanged from Muse Spark 1.2: $1.25/$4.25 per 1M input/output tokens, with cache hits discounted to $0.15 per 1M
➤ Input modalities: text, image, video
➤ Availability: Meta's first-party API and Muse Code
best part about coding being solved is no one is making the “debugging for hours cause i missed a semicolon” joke anymore.
one of a million things i love about nvidia is how cool their GPUs look. they would sell just as well even if they didn’t look pretty but the fact that they do makes me think someone put some effort into it and i appreciate that.
for the first time in a while, i don’t have an impending deadline. i also temporarily limited myself to 1 meeting a day and deprioritized replying to texts. it felt weird at first, but it’s brought me so much mental clarity.
it’s like ive been collecting random data points for the last 4 years but never got a second to sit down and process it. its evocative of the feeling i got when i used to play football and id take off my helmet after a really long practice.
im still working a lot of hours but most of that time is spent just sitting in front of a white board thinking. the problem i need to spend the next years of my life working on has become increasingly clear.
I recommend it to anyone who’s been heads-down for years. doesn’t have to be a month but take a week or so to just sit and process.
at the age of 22, ive discovered how fun legos are teapawt.net/i/jf/cards/money/share…
Raman Khatri retweeted
The next few years in SF will be incredible.
Many big lab employees will leave . The idea of sailing into the sunset is temporal and will quickly subside. Inevitably, they will start their own companies and absorb a tremendous amount of venture capital. This will lead to an abundance of innovation.
On the negative side, a lot of this innovation will be in the largely overlapping homogenous areas where their skillets lie. We see this already: many companies doing AI for quant trading, drug discovery, material discovery, robotics, new AI architectures, hardware. Competition in such disciplines will be severe.
Yet on the positive side, never in history could you raise this quantum ($100M+) of risk capital out of the gate and the space of invention that begets is unprecedented. People will take swings that were once not viewed as VC backable businesses. And yes, beyond just buying loads of compute to train models. Moonshot ideas which usually only worked under the comfort of a big company. Instead of 100, there may be 1000 startups dubbed “neolabs”.
Many will invest capital of their own. They will all pay top dollar to attract talent. They will poach large swathes of academia out of their labs to help.
Many will not find the hunger to succeed and get lazy from their riches. The more successful ones will acquire them if they are colleagues they liked.
The amazing ones will succeed immeasurably. And this will beckon a new renaissance we won’t be able to fathom from where we are now.