Ilya retweeted
Pretty wild the woman in this video is a realtime generated agent
The whole point of building Tavus has been simple: talking to a machine should feel as natural as talking to a friend or coworker. It’s hard to describe all the tiny nuances that make a conversation feel human. The little expressions. Moving around in your chair. Knowing when to speak and when to listen. The dance of it all. Griffin is by far the closest anyone has come to a model that can capture those nuances. The first time I saw it being used, I had no idea I was watching our model rather than just a normal video call. I’m so incredibly proud of this team and what they’ve built.
263
221
3,639
717,127
Yaniv, the Israeli passenger who neutralized the co-pilot aboard the flydubai flight to Israel, gave his account to Prime Minister Benjamin Netanyahu at Ben Gurion Airport after the plane landed safely. The cockpit door was jammed and the co-pilot was trying to destroy the aircraft's instruments. Yaniv put him in a chokehold. Other Israeli passengers helped drag him out. Yaniv then went back to the controls. "I pulled the yoke up and stopped the plane's dive. Lucky I watched Air Crash Investigation."
11
25
449
45,432
RT @Osint613: The Israeli passenger who subdued the Omani co pilot trying to crash the Flydubai flight carrying about 180 passengers, inclu…
1,137
1
A meeting on Iran with senior officials from several Gulf states was held during Netanyahu's visit to the UAE, some of them from countries Israel has no formal relations with. Government and military aircraft landed in Abu Dhabi alongside the visit: a Libyan government plane, a Moroccan Hercules, a Qatari military plane, and an Emirati government plane flown in from Riyadh. Kan.
1
4
36
25,752
It’s the producers who eat the freight cost, not the refiners. Cost of freight has no impact on landed oil price until it’s high enough to force the producers to decrease oil output.
Chaps, my data and my message below are very clear: Iran has lost control of Hormuz. But careful with over-reading my message. I never said we have an oil bear market now. We are not there yet. For one, the Hormuz Shuttle needs to become more permanent and more efficient. The Hormuz Shuttle creates enormous tanker inefficiencies, driving up landed costs for refineries, especially in Asia. Just as importantly, product exports remain heavily disrupted because of Russia, the East–West pipeline, tanker inefficiencies and lower Chinese refinery runs. That means cracks remain high — and diesel and gasoline prices remain high — even if crude falls below $100. Next problem: the moment China starts buying nearer its 9–10mbpd seaborne import baseline, we still lack sufficient crude flows and face even greater tanker-rate stress, pushing both crude & product prices higher again. Above all, Trump needs to stay the course if this regime is to break and flows are finally to normalise. If he goes for another Memorandum of Misunderstanding, all he does is hand the regime another lifeline and allow it to rinse and repeat the Q3 chapter. Patience. This is not yet a deflationary signal for an equity rally. But we are making progress. Come Q1 2027, assuming Trump holds the line, this should get really interesting. Once flows normalise and tanker inefficiencies ease, the market will front-run the balances and push crude significantly lower. The UAE will push production towards 5mbpd. What little remains of OPEC will produce flat out to regain market share. We may rediscover just how weak underlying Chinese crude consumption really is. Perhaps even an AI investment bust? All bearish. Q2 2027? IDK. But then we still need to refill some 1.5–2.0bn bbl of crude stocks lost globally. How quickly that happens — and at what price — will matter enormously. Lots of moving parts. Let’s keep an open mind. @LSREnergy
24
Sinokor is tankers market’s OPEC. “The Sinokor effect is still in play. They have a huge number of vessels, which they prefer to keep idle rather than let the market fall. This is continuously adding to inefficiency.”
1
1
31
🛢️ OIL CONTEXT WEEKLY 🛢️ Saudi Arabia’s East-West Pipeline entirely offline this week and will remain throttled for a month or more, pushing Riyadh headlong into the Hormuz dark trade, while diesel prices rose above $220/bbl. Summary below, full report in reply.
15
22
136
59,771
Replying to @CRUDEOIL231
8
16
185
21,261
So true. Been waiting for it since 90s. Happened more or less when I expected it to happen. Real and urgent existential risk for all people.
I resigned from Anthropic today. I spent the last three years doing pretraining research at both OpenAI and Anthropic. Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives. More thoughts below.
1
51
SPR weekly releases: 6.1 → 5.3 → 3.7 → 3.1 → 1.2 Trump is out of oil as he notified upon signing the MoU
26
Ilya retweeted
tristan buckmaster just published a statement that is, if accurate, one of the ugliest things i’ve read out of a frontier lab. not because of the math. because of what happened after openai found out two people were about to beat them to the punch. buckmaster and levent alpöge spent a year pushing the córdoba / martínez-zoroa program to smooth forcing. august 15 they got blowup for boussinesq and 3d euler. lean-verified august 22. they were paying openai out of pocket and dumping every draft into private codex sessions. the program is obscure. almost nobody else was on it. this was not “paste the clay problem into a chatbot.” then this: > thursday sept 3, buckmaster emails openai privately. rumors are flying. he is trying to stop a mess, not start one. > they reply the same day: give us details so we can “avoid competing.” also, want free compute? > sunday they suddenly need a call “at any point today.” sebastien bubeck gets on. levent is not invited. > openai says an internal model has a ~100-page proof of forced navier–stokes blowup. fefferman options c and d. the exact route buckmaster and alpöge had quietly chosen. > they show him a prompt and claim the model was “simply given the problem statement.” levent is told “very little human input.” > that falls apart on the same call. a whole team had been on it. they started on the unforced problem, warmed the model up on easier equations including euler, and even the prompt they showed him was written by prompting codex. > he asks when the first prompt went out. they stall. then agree: the past few days. after information about his work reached openai. > he asks whether the model was trained on, or had access to, the private codex sessions where they put every draft of this project. > answer: the model does not look up user data. > he asks again. about training. > no answer. then it stops being a science story and becomes an hr story. > offer 1: you post euler. we post navier–stokes the next day. > offer 2: after you post euler, you alone write the navier–stokes paper, credit our model, and we cut levent. > bubeck says twice he wants levent removed from authorship because he works at anthropic. he declines both. he says if they release it that way, he goes public. the reply is not a scientific objection. > “why would you ruin your career?” > “if you don’t want me to be nice, then i don’t have to be nice.” later, a text to levent proposing a one-on-one: > “i don’t know if tristan is being fully rational right now.” that is not how you treat a collaborator. that is how you split one. openai got word two researchers were closing a path almost nobody else was on. they spun up a team on that exact path. they would not answer whether the researchers’ unpublished drafts in openai’s product were used. they tried to dictate the announcement order. they tried to remove a coauthor because he works at a competitor. when that failed, someone reached for a career threat. this is a lab finding out academics were about to publish, sprinting the same narrow program, refusing the data question, and trying to manage credit around a rival affiliation. really slimy if even half of this holds. what the fuck is happening.
holy shit. tristan buckmaster is saying openai has a forced navier–stokes blowup via the córdoba / martínez-zoroa program. and that it may have used drafts from his and levent alpöge’s private codex sessions. that’s the córdoba / martínez-zoroa route. almost nobody else was on it. he also says the first prompt went out after information about his and levent alpöge’s work reached openai. that “very little human input” was not true. that a whole team had already been set on easier problems, including euler. that even the prompt they showed him was written by prompting codex. then he asked: did the model have their private codex sessions. no answer on training. if any of that holds, openai didn’t just scoop a math result. they may have done it off two researchers’ unpublished drafts, then tried to cut the coauthor because he works at anthropic. a millennium-problem-adjacent result just became employee drama before an ipo. i don’t even know which part is crazier. this is one of the slimiest things i’ve seen in this industry.
234
2,415
15,197
1,810,289
Houses? There are 184 Iran-tankers which OFAC (part of the US Department of Treasury) still haven't slapped with sanctions. They alone have transported 533 million barrels (let's say $32 billion) of Iranian crude oil and refined products over the years.
U.S. Treasury Secretary Scott Bessent, to the Iranian regime leadership: "I will say it on worldwide TV: we know where in the British Virgin Islands your accounts are, at these trust companies. We know the hundred million dollar houses you have around the world. we are going to close all that down." "We are going after the regime's assets they have stolen from the Iranian people." "They can go to the victims of terror." "After 47 years -- it is over." "Either you are with us, or against us."
17
71
459
91,716
ACTIVE FUND MANAGERS ARE LOSING THE LONG-TERM BATTLE TO PASSIVE Only 27% of actively managed U.S. large-cap funds survived and beat their passive peers in the 12 months through June 2026. Over 10 years, only 13% managed to outperform. That means roughly 87% failed to beat passive alternatives over the decade.
1
11
59
28,999
Bessent bled his fund to death being short oil in 2003-04. This is starting to sound like a revenge trade.
Ok, fine, let’s do this. I knew of Bessent when he worked in the London office at Soros & heard pitches for each one of his subsequent ventures listed below & for his last venture, a FoF w/ Jeff Tarrant, not included below. In short summary, his claimed valor for the famed British pound cross trade is a myth. Soros managed his own principal positions & risk…always. There were other “investing teams” that also traded currency crosses, including the one Bessent was a part of in London, but to be clear that was a semi-autonomous group w/ its own capital & risk budgets allocated by Soros internally. They had nothing - NOTHING - to do w/ George’s massive bet against BoE interventions. Full stop. Like many who stayed in the business after Soros wound up in 2000-early 2001, Bessent’s team at Soros broke off to do their own fund w/ a capital allocation from Soros’ family office. They bled to death being short oil & long US equity volatility in 2003-04, which introduces my point about Bessent which has always held up through the years. Scott Bessent is a contrarian. Being a contrarian caused his group to bleed out at Soros in early 2000, his first fund to fail in the mid 2000s, his second fund to fail in the mid-to-late 2010s (at one point he took the other side of the yen carry trade in 2017…really). And so on… The problem with contrarians is the dogma “I’m not wrong, I’m just early” which leads to hubris “market still wrong, have to stay with it”…until the lights dim & losses compound to the point where the business is no longer viable. Scott Bessent did that death spiral THREE times before becoming a marketing guy at a FoF. For a career trader that’s like going from starting for the New York Yankees to being the Kenny Powers tv series character trying to make it happen on a low rung minor league team in Myrtle Beach. It’s glib to criticize his experience: that’s not the real issue now. The real issue is Bessent is a serial contrarian & that is the worst bad thing if you are a financial speculator in the markets. That’s it. @RickPetree @NastyOldWomynII @GordonJohnson19
15
108
1,816
138,240
Something crazy and unique about this diesel rally is that it isn't just at the front of the curve. Here's the US diesel futures curve today (white) vs at the peak in March (blue) Far broader rally across the curve
18
33
227
21,485
"Compute is the new oil" ~ OpenAI President Greg Brockman We call this the pipeline treatment. Polling via @heatmap_news
7
6
98
11,636
PSA: China is back buying. #oott
8 mb of Basrah Medium and Basrah Heavy cargoes just sold to Chinese refiners. Saudi allocations have edged up to 14 mb. Russian ESPO is sold out for October. @JuneGoh_Sparta on why this could be China's crude demand inflection point. Read the full crude analysis here: spartacommodities.com/market… #oilmarkets #oott #commoditytrading
4
32
176
22,655
Ilya retweeted
So we keep talking about Ukrainian drones attacking Russian refineries. Pulling out my Refinery Configuration Schematic, I see 3 big scenarios: 1⃣If the drones struck the CDU (the crosses in blue), there's refinery downtime and the immediate hit is on distillates production and more crude exports. 2⃣If the drones struck the reformer (the small yellow crosses), you are going to get less mogas but more naphtha exports from Russia. 3⃣If the drones struck the important secondary units like the FCC and HCU (the big red crosses), you are going to get less mogas and diesel respectively, and more fuel oil exports from Russia (if the CDU is still running). Also it usually takes a longer time to repair damaged secondary units vs the CDU due to more sophisticated parts. I don't have visibility of where the strikes happened, but we can see European product cracks for direction. #oott
5
23
106
36,799