Offshore
Photo
The Transcript
Zillow CEO: "We delivered strong results across the business and achieved all of our reported financial targets for the full year — including full-year profitability — and we’re carrying that momentum into 2026."
$Z: -5% AH https://t.co/6WZHqivy3b
tweet
Zillow CEO: "We delivered strong results across the business and achieved all of our reported financial targets for the full year — including full-year profitability — and we’re carrying that momentum into 2026."
$Z: -5% AH https://t.co/6WZHqivy3b
tweet
Offshore
Photo
The Transcript
Klaviyo CFO: "Our results reflect strong execution and growing demand, with 32% annual revenue growth, expanding operating margins, and strong cash flow"
$KVYO: +7%AH https://t.co/1SLlzPs2g5
tweet
Klaviyo CFO: "Our results reflect strong execution and growing demand, with 32% annual revenue growth, expanding operating margins, and strong cash flow"
$KVYO: +7%AH https://t.co/1SLlzPs2g5
tweet
Offshore
Photo
God of Prompt
RT @godofprompt: 🚨 Holy shit… Stanford just published the most uncomfortable paper on LLM reasoning I’ve read in a long time.
This isn’t a flashy new model or a leaderboard win. It’s a systematic teardown of how and why large language models keep failing at reasoning even when benchmarks say they’re doing great.
The paper does one very smart thing upfront: it introduces a clean taxonomy instead of more anecdotes. The authors split reasoning into non-embodied and embodied.
Non-embodied reasoning is what most benchmarks test and it’s further divided into informal reasoning (intuition, social judgment, commonsense heuristics) and formal reasoning (logic, math, code, symbolic manipulation).
Embodied reasoning is where models must reason about the physical world, space, causality, and action under real constraints.
Across all three, the same failure patterns keep showing up.
> First are fundamental failures baked into current architectures. Models generate answers that look coherent but collapse under light logical pressure. They shortcut, pattern-match, or hallucinate steps instead of executing a consistent reasoning process.
> Second are application-specific failures. A model that looks strong on math benchmarks can quietly fall apart in scientific reasoning, planning, or multi-step decision making. Performance does not transfer nearly as well as leaderboards imply.
> Third are robustness failures. Tiny changes in wording, ordering, or context can flip an answer entirely. The reasoning wasn’t stable to begin with; it just happened to work for that phrasing.
One of the most disturbing findings is how often models produce unfaithful reasoning. They give the correct final answer while providing explanations that are logically wrong, incomplete, or fabricated.
This is worse than being wrong, because it trains users to trust explanations that don’t correspond to the actual decision process.
Embodied reasoning is where things really fall apart. LLMs systematically fail at physical commonsense, spatial reasoning, and basic physics because they have no grounded experience.
Even in text-only settings, as soon as a task implicitly depends on real-world dynamics, failures become predictable and repeatable.
The authors don’t just criticize. They outline mitigation paths: inference-time scaling, analogical memory, external verification, and evaluations that deliberately inject known failure cases instead of optimizing for leaderboard performance.
But they’re very clear that none of these are silver bullets yet.
The takeaway isn’t that LLMs can’t reason.
It’s more uncomfortable than that.
LLMs reason just enough to sound convincing, but not enough to be reliable.
And unless we start measuring how models fail not just how often they succeed we’ll keep deploying systems that pass benchmarks, fail silently in production, and explain themselves with total confidence while doing the wrong thing.
That’s the real warning shot in this paper.
Paper: Large Language Model Reasoning Failures
tweet
RT @godofprompt: 🚨 Holy shit… Stanford just published the most uncomfortable paper on LLM reasoning I’ve read in a long time.
This isn’t a flashy new model or a leaderboard win. It’s a systematic teardown of how and why large language models keep failing at reasoning even when benchmarks say they’re doing great.
The paper does one very smart thing upfront: it introduces a clean taxonomy instead of more anecdotes. The authors split reasoning into non-embodied and embodied.
Non-embodied reasoning is what most benchmarks test and it’s further divided into informal reasoning (intuition, social judgment, commonsense heuristics) and formal reasoning (logic, math, code, symbolic manipulation).
Embodied reasoning is where models must reason about the physical world, space, causality, and action under real constraints.
Across all three, the same failure patterns keep showing up.
> First are fundamental failures baked into current architectures. Models generate answers that look coherent but collapse under light logical pressure. They shortcut, pattern-match, or hallucinate steps instead of executing a consistent reasoning process.
> Second are application-specific failures. A model that looks strong on math benchmarks can quietly fall apart in scientific reasoning, planning, or multi-step decision making. Performance does not transfer nearly as well as leaderboards imply.
> Third are robustness failures. Tiny changes in wording, ordering, or context can flip an answer entirely. The reasoning wasn’t stable to begin with; it just happened to work for that phrasing.
One of the most disturbing findings is how often models produce unfaithful reasoning. They give the correct final answer while providing explanations that are logically wrong, incomplete, or fabricated.
This is worse than being wrong, because it trains users to trust explanations that don’t correspond to the actual decision process.
Embodied reasoning is where things really fall apart. LLMs systematically fail at physical commonsense, spatial reasoning, and basic physics because they have no grounded experience.
Even in text-only settings, as soon as a task implicitly depends on real-world dynamics, failures become predictable and repeatable.
The authors don’t just criticize. They outline mitigation paths: inference-time scaling, analogical memory, external verification, and evaluations that deliberately inject known failure cases instead of optimizing for leaderboard performance.
But they’re very clear that none of these are silver bullets yet.
The takeaway isn’t that LLMs can’t reason.
It’s more uncomfortable than that.
LLMs reason just enough to sound convincing, but not enough to be reliable.
And unless we start measuring how models fail not just how often they succeed we’ll keep deploying systems that pass benchmarks, fail silently in production, and explain themselves with total confidence while doing the wrong thing.
That’s the real warning shot in this paper.
Paper: Large Language Model Reasoning Failures
tweet
Offshore
Photo
The Transcript
LYFT CFO: "We delivered record financial performance in 2025 across all metrics, including all-time-high cash flow generation exceeding $1.1 billion...we remain right on track to hit our long-term targets."
$LYFT: -15% AH https://t.co/JNjkPeJDK9
tweet
LYFT CFO: "We delivered record financial performance in 2025 across all metrics, including all-time-high cash flow generation exceeding $1.1 billion...we remain right on track to hit our long-term targets."
$LYFT: -15% AH https://t.co/JNjkPeJDK9
tweet
Offshore
Photo
God of Prompt
RT @alex_prompter: 🚨 The guy who built Anthropic’s defenses against AI bioterrorism just quit.
Mrinank Sharma led Anthropic’s Safeguards Research Team. His job was literally making sure Claude doesn’t help bad actors do bad things.
His resignation letter: “The world is in peril. And not just from AI, or bioweapons, but from a whole series of interconnected crises.”
He also said he “repeatedly seen how hard it is to truly let our values govern our actions” inside the organization.
This is the company that positioned itself as the “safe” AI lab. The one founded specifically because OpenAI wasn’t careful enough.
Now their safety lead is walking away, saying the pressure to “set aside what matters most” is real.
He’s leaving to study poetry. Not joining a competitor. Not starting a startup. Poetry.
When your AI safety researcher chooses poems over production, that tells you something about what’s happening behind closed doors.
tweet
RT @alex_prompter: 🚨 The guy who built Anthropic’s defenses against AI bioterrorism just quit.
Mrinank Sharma led Anthropic’s Safeguards Research Team. His job was literally making sure Claude doesn’t help bad actors do bad things.
His resignation letter: “The world is in peril. And not just from AI, or bioweapons, but from a whole series of interconnected crises.”
He also said he “repeatedly seen how hard it is to truly let our values govern our actions” inside the organization.
This is the company that positioned itself as the “safe” AI lab. The one founded specifically because OpenAI wasn’t careful enough.
Now their safety lead is walking away, saying the pressure to “set aside what matters most” is real.
He’s leaving to study poetry. Not joining a competitor. Not starting a startup. Poetry.
When your AI safety researcher chooses poems over production, that tells you something about what’s happening behind closed doors.
Today is my last day at Anthropic. I resigned.
Here is the letter I shared with my colleagues, explaining my decision. https://t.co/Qe4QyAFmxL tweet
Offshore
Photo
The Transcript
Ford CEO: "..a strong 2025 in a dynamic and often volatile environment."
$F: +1.5% AH https://t.co/DGOZHaa76H
tweet
Ford CEO: "..a strong 2025 in a dynamic and often volatile environment."
$F: +1.5% AH https://t.co/DGOZHaa76H
tweet
Offshore
Photo
Fiscal.ai
Cloudflare just added $353 million in new Remaining Performance Obligations.
That's their largest quarterly increase ever.
$NET https://t.co/ysJMNkpFgg
tweet
Cloudflare just added $353 million in new Remaining Performance Obligations.
That's their largest quarterly increase ever.
$NET https://t.co/ysJMNkpFgg
tweet
Offshore
Photo
App Economy Insights
RT @EconomyApp: A quick look at the memory crunch:
📲 $QCOM: Memory Wall
🎮 $SONY: Hardware Retreat
☁️ $ARM: Cloud AI Engine Ignites
Full breakdown of the AI ripple effect.
https://t.co/XthMtTFjEX
tweet
RT @EconomyApp: A quick look at the memory crunch:
📲 $QCOM: Memory Wall
🎮 $SONY: Hardware Retreat
☁️ $ARM: Cloud AI Engine Ignites
Full breakdown of the AI ripple effect.
https://t.co/XthMtTFjEX
tweet
Offshore
Video
God of Prompt
RT @rryssf_: 🦞 OpenClaw has 114,000+ GitHub stars and the whole tech world is losing its mind over it.
But here's what nobody's showing you: the setup process that made 90% of people quit before their first agent sent a single message.
Node.js configs, gateway daemons, Tailscale tunnels, security hardening...
There's now a way to skip all of it.
Here's what i found:
A plug-and-play approach that turns autonomous agents into something you can spin up in minutes and wire into any API you want.
https://t.co/FDCVmoBR3o
For say a daily ai newsletter:
tweet
RT @rryssf_: 🦞 OpenClaw has 114,000+ GitHub stars and the whole tech world is losing its mind over it.
But here's what nobody's showing you: the setup process that made 90% of people quit before their first agent sent a single message.
Node.js configs, gateway daemons, Tailscale tunnels, security hardening...
There's now a way to skip all of it.
Here's what i found:
A plug-and-play approach that turns autonomous agents into something you can spin up in minutes and wire into any API you want.
https://t.co/FDCVmoBR3o
For say a daily ai newsletter:
tweet