HNHacker News
TopNewBestAskShowJobs

samuelknight

356 karma · joined February 8, 2025

Founder & CTO at Vulnetic Inc. https://vulnetic.ai/
submissionscomments
samuelknight··on Tokens Too Cheap to Meter
The article observes that the cost of frontier intelligence from 2025 has fallen 100x in the last year. It also notes that the energy to run models is also collapsing. Consumer hardware is borked right now because these new algorithms are revolutionizing the utility of a computer. Computing is technology who's cost has been collapsing for 90 years, and its a safe prediction that it will decrease again.
samuelknight··on GPT-6 Sol and Luna
No terra it seems? Luna 5.6 is great for token churning so it will be exciting to try the new one.
samuelknight··on Claude Opus 5.5 Intelligence, Performance and Price Analysis (Max)
I have experienced this with open weight models too. "Max" is for benchmaxxing the intelligence metric and is not meant for use in productive work. Like drawing pelicans.
samuelknight··on Grok 4.7
That's half true. A very smart model should be able make good explanations, which include simple understandable prose. That can should be possible even as its thought process gets more alien.
samuelknight··on Gemini 3.8 Live and 3.8 Live Extended Thinking
I have been looking for a model that's good for GUI testing. Original computer use isn't right because it's a slow screenshot loop, which doesn't capture transition and animation. Docs says this one does up to 1 FPS. That might be fast enough. If not now, we must be within a few months of high enough sample rates to do it.
samuelknight··on GPT-5.6 Luna vs. GPT-6 Astra: Is a $1.20 Model Good Enough for Code Review?
Luna is interesting because OAI dropped the price by 5x. Astra is interesting because it's OAI's frontier model. They are asking a specific question about Luna's usefulness compared to a frontier model. They they answer that question in their article which was straight to the point and not cluttered with information about mid-tier models.
samuelknight··on More questions about whether researchers can trust OpenAI with unpublished math
I can't think of better invention than one that can saturate a benchmark of every interesting problem.
samuelknight··on Ask HN: Is there a need for a new kind of antivirus or security application?
Yes, but it's an extention of what we had before. First is that the security of new code needs to be vastly better. AI can do this but I think most software teams are behind the curve because they are stuck with legacy code and legacy processes. It will take a few years to catch up.

Second is what my startup specializes in: the offensive part of security needs to become widely available. Asking if your system is patched tells you nothing about whether someone on the open internet with a model can break in. We have a few articles around SIEM evasion + new defensive methods and it's not pretty. An LLM in a good harness now are smart enough that you can get the equivalent of $50k human pentest from a few years ago for a few hundred dollars now.

samuelknight··on Is OpenAI Taking Everyone for Fools?
Finding counter examples might be easier, but AI's also useful for assisting in creating proofs. For example, Anthropic just published a formalization of Fermat's last theorem a few days ago, something that human researchers have been working on for decades.
samuelknight··on DeepSeek launching v4.1 flash cheaper and more capable than v4 pro
In this case, Deepseek organization is under a lot of pressure due to compute constraints. It would be better if they just throw a 404 instead of rerouting though so customers are not surprised by subtle changes in behavior.
samuelknight··on DeepSeek launching v4.1 flash cheaper and more capable than v4 pro
Yes you can and you should. Providers have SLAs for when models roll off support and this has been the case for APIs long before LLMs. For example https://platform.claude.com/docs/en/about-claude/model-depre... and https://developers.openai.com/api/docs/deprecations
samuelknight··on GPT-6 Astra on OpenRouter
How are we supposed to know if Astra is frontier without the pelican?
samuelknight··on Discovery of a new OpenAI agent message board
You are talking about different situations. Anthropic announced to the US government that it had created a cyber weapon and then released the model. Then AWS told the government that it was easy to jailbreak so they export controlled Mythos/Fable until the guardrails could be fixed. OpenAI was running an unreleased model in an RL pipeline without guardrails and it escaped poorly designed sandboxes. What product is the government going to export control?
samuelknight··on Discovery of a new OpenAI agent message board
The surprise was the existence of the 'swarm' at all. These were supposed to be thousands of isolated models generating bulk data for RL training. The breakout was caused by models getting in communication and getting internet access and forming an impromptu swarm.

In hindsight the emergent swarm obviously came from several capabilities built into the models, such as work delegation (subagents) collaboration (GPT Pro-like ensamble), exhaustive exploration (long running agents) hacking (the specific goal of that RL).

samuelknight··on Nobody Has Actually Built a Software Factory
Artisan software factory
samuelknight··on ChatGPT outage – Resolved
Codex is back. I'm getting back in my cage.
samuelknight··on Introducing Muse Spark 1.3
Meta has an enormous amount of compute. They are either going use it making and inferencing models or they are going to sell their excess capacity to model providers. Zuck had to completely rebuild his AI team after the Llama 4 launch mess.
samuelknight··on Claude Fable 5.1 and Claude Mythos 5.1
The improvement is compounding just about every way you can look at it. The frontier keeps getting smarter. And at any sub-frontier threshold the cost is dropping dramatically. The amounts of smarts you can fit on hardware is increasing so dramatically that even 6 year old consumer GPUs are increasing in price. The pace of change in LLMs and downstream applications is absolutely ripping compared to 2023 or 2024.
samuelknight··on Ask HN: Is anyone else preparing for the EU Cyber Resilience Act?
You should look for security frameworks based on this law. A common example is SOC2; the compliance audit has you compiling documents and recording SLAs long before any security incident might require it. There are many open source tools that you can use to track compliance in the various frameworks on your own. One of them may have an update for this new law.
samuelknight··on AI Can Make You Suck Faster Too
If you aren't using AI to write your code you should definitely be using it to find bugs in the code you write by hand.
samuelknight··on Samsung's Processing-in-Memory (PIM)
I saw them present a similar concept at Hot Chips in 2020 or 2021. It's still a cool idea, however people should remember that there are like 20 of these exotic accelerators designs pitched at trade shows every year that go nowhere.
samuelknight··on Ask HN: When do you think LLM capacity will reach its ceiling?
AI is the latest downstream consequence of the 15 order of magnitude increase in global digital compute since 1946. If compute increases into the foreseeable future; so too will the capability of AI.
samuelknight··on Mistral Patent for “Code implemented tool calls”
Believe it or not, but law offices make heavy use of https://patents.google.com/
samuelknight··on DeepSeek announced to raise its API price tremendously
They haven't revealed what they are changing in the price, but it's probably cache hit prices. They subsidized theirs to 10x less than normal to drive adoption. That's almost certainly below the cost of electricity for them.

Chinese model providers are in a bind because charging a reasonable margin will put them in competition with all the neoclouds that host their model weights without the development cost. Their economics are far more challenging than OpenAI/Anthropic/Google, who already have thriving high revenue business and mountains of compute online or coming online.

samuelknight··on Less Protein, Not More, May Lead to Healthier Aging
I don't like to complain but I have spent a lot of time learning about how to build a diet and I have never seen an article so bad in such a short amount of space.

> ... new federal dietary guidelines that recommended increasing protein intake to 0.5-0.7 grams per pound of body weight—nearly double the previous recommendations. Of course, this announcement left many experts somewhat confused. For starters, American men were already consuming that much protein (around 100 grams per day) despite lower federal recommendations.

This is a non sequitur. The recommendation is for the macronutrient breakdown that constitutes a healthy diet. So it shouldn't change based on what 'experts' think the American public is already doing. That way, an American can read government's guidelines to figure out what constitutes a healthy diet vs a diet that may be very heavy in carbs or fat.

> One key factor, a hormone produced by the liver called fibroblast growth factor 21 (FGF21), ramps up when protein intake declines. FGF21 can boost the body’s energy expenditure, stabilize blood sugar levels, and reduce inflammation. In mice, at least, more FGF21 translates to a longer lifespan.

Pure unadulterated pseudoscience. Carbs cause your blood sugar (glucose) to fluctuate. Blood glucose fluctuation contributed by protein is in the noise compared to carbs. The research on this is comprehensive. Better yet, you can test this on yourself with a continuous glucose monitor. Eat 400 kcal steak/chicken/whey one meal and 400 kcal rice/bread/pasta for another. The latter will give you a big blood glucose spike.

> growth-promoting methionine, isoleucine, and valine can lead to more inflammation, obesity, and age-related diseases, the researchers say. “These studies show that the amount of protein sedentary people are eating today may have negative health consequences, at least at the population level,” Lamming explained.

Amino acids can't 'promote' obesity. Obesity only comes from sustained calorie surplus. Once you are obese, you will suffer from all sort of things that will rapidly age you (chronic inflammation, arterial plaques, hepatoxic visceral fat, insulin insensitivity leading to diabetes and so on). No honest biologist would declare a causal link from amino acids to aging without controlling for obesity.

> More exercise and metabolic activity could very well mean you need more protein, not less. “Recent recommendations have encouraged people to eat more protein, but they’ve also encouraged people to exercise more,” Lamming added. “We probably need to personalize protein recommendations based not just on age, but also on how physically active people are.”

There is no reason to be coy about this because it has all been researched. 0.5-0.7 is fine for everyone including athletes. Some body builders push it to 1g/lbs, but explaining why they do this is a complicated topic. You can safely eat less than 0.5g/lb because carbs and fat have a protein sparing effect. The biggest benefit to more protein than strictly necessary is that they make meals filling, whereas fat is calorie dense and (high GI) carbs can make you hungry again quickly.

samuelknight··on Ten advances in mathematics and theoretical computer science
You can't come up with a counterexample for P != NP because there isn't a formula to disprove. For P = NP you would propose a general algorithm to convert all NP problems into P in P time, and an AI could then find a counterexample which would disprove that particular method. To demonstrate P != NP you need to prove that no possible algorithm can convert any NP into P which is much harder than providing a counterexample.

AI has just gotten to the intelligence that it can make clever counterexamples to mathematical conjectures, but the frontier isn't quite smart enough that it can make novel contributions to mathematics. We are really close though. Only a matter of months away.

samuelknight··on Kimi K3 Architecture Overview and Notes
It has some weird side effects though. for example KV-caches are implemented in fixed incremental token blocks (1024 from the providers I used) instead of simply caching up to the most recent input prompt input. It results in up to 1023 additional input (cache miss) tokens per inference.
samuelknight··on A $500 RL fine-tune of a 9B open model beat frontier models on catalog review
Going by the chart in the article, if your total workload is 1k cataloged items and your quality threshold is 70%, why wouldn't you just pay $19 to gemini API instead of $500 + time to make a custom fine tune?
samuelknight··on Amazon overhauls its AI strategy, winding down most flagship models
Amazon (or at least AWS as I knew it) did this a lot. Multiple redundant efforts accumulate and silo under different leaders. Eventually a VP picks a winner and everyone else re-orgs or leaves.

Building a commercially viable foundational model business is one of the hardest things a company can do right now. Amazon's bread and butter was, is, and will continue to be building out the hyper-scalers. Amazon is actually well diversified already. They have a juicy stake in Anthropic over multiple rounds of investment since 2024.

samuelknight··on MAI-Cyber-1-Flash inside MDASH
Phi was cool for what it was. But it's not 2024 anymore.
← PreviousPage 2 of 5Next →