
building in ai, web & neurotech! founding ai eng @propane_ai 🇩🇰 | prev: sr ai eng @neuronsinc 🇩🇰 comp neuro @imperialcollege 🇬🇧 cs @unibirmingham
LDN🇬🇧/KBH🇩🇰
Joined October 2021
- Tweets3.8K
- Following1.2K
- Followers566
- Likes15.7K
Ben Fleming retweeted
In a weird way, everybody is rooting for you. When you go to a job interview, the interviewer wants you to be amazing so they don’t have to look anymore. when you go on a date, the person wants to have an amazing date with you. If you make a piece of content, people want to see amazing pieces of content and they have no problem if it’s yours.
the instant you get into the mindset that people want you to do well, your entire world gets better. And you’ll be shocked to know this pattern keeps repeating in different areas of life.
Ben Fleming retweeted
Dusting off this tweet from April because this gap in shared understanding of LLM capability is *widening*. It's now less "two groups of people speaking past each other" and more a sharp funnel.
- Napkin math somewhere around 6B people (~75% of the population) have barely come in contact with LLMs at all.
- Around 1-2B (~20%) are casual and infrequent users of free-tier ChatGPT-like products. This group sees derpy chatbots and treats them a bit like a better Google search, a writing aid, or etc. Maybe an agent tries to book you a flight. I have non-tech friends who (reasonably, imo) say they have not much use for it at all. Even many professionals outside of math&code are in this tier. For example, execs and many other functions spend a lot of their time talking to other people, so while they understand what is happening intellectually, it is still second-hand and a bit abstract.
- Now we get to professional use of frontier-grade LLMs in math&code. Somewhere around 20M people (0.2%) see first-hand that large, complex projects that used to take them weeks/months can now be completed by agents with a prompt. Building apps, copying apps, translating apps, decompiling apps from binaries... This has all happened very quickly and recently - less than 1 year ago, I was writing code manually by hand, typing memorized computer code commands into a code editor character by character, occasionally pressing Tab to autocomplete a little chunk of code.
- And finally we get to the ~5,000 people (~0.00006%) with access to frontier-grade systems internally. The external world has seen the preview. It looks like swarms of thousands of agents collaborating over weeks on software mega projects: minting zero days, running cyber attacks and defenses at machine speeds, discovering new science, advancing the frontier of mathematics. Things that would have taken top professionals in the industry years of work. Meanwhile, human review and comprehension are starting to fall behind. For example, people are still involved in the "archeology" of the OpenAI-HF incident from many months ago. Mathematicians may be poring over the 722 manuscripts on frontier mathematics for a while.
The funnel is driven by a combination of factors:
1. The impact scales with ambition, problem size, and horizon. A question with a paragraph answer barely stresses the system. You need a reservoir of big, difficult problems that you really care about. This aspect drives the consumer / professional dimension of the funnel.
2. The jaggedness of the system (which I have written about a lot separately). Capability peaks in domains that are digital, verifiable and economically valuable. This is because LLM capability emerges from reinforcement learning on verifiable rewards on a curated environment mixture driven by revenue potential. This aspect primarily drives the area (e.g. math&code) dimension of the funnel.
3. Access. Free-tier, paid-tier, internal.
So this is the weirdness of the moment. The general public has mostly not interacted with these systems. When they have, it looks like a derpy chatbot. The majority of professionals still see only a modest uplift. And a small sliver of professionals are experiencing the vertigo of the curve going vertical. And it is all happening at the same time.
Judging by my tl there is a growing gap in understanding of AI capability.
The first issue I think is around recency and tier of use. I think a lot of people tried the free tier of ChatGPT somewhere last year and allowed it to inform their views on AI a little too much. This is a group of reactions laughing at various quirks of the models, hallucinations, etc. Yes I also saw the viral videos of OpenAI's Advanced Voice mode fumbling simple queries like "should I drive or walk to the carwash". The thing is that these free and old/deprecated models don't reflect the capability in the latest round of state of the art agentic models of this year, especially OpenAI Codex and Claude Code.
But that brings me to the second issue. Even if people paid $200/month to use the state of the art models, a lot of the capabilities are relatively "peaky" in highly technical areas. Typical queries around search, writing, advice, etc. are *not* the domain that has made the most noticeable and dramatic strides in capability. Partly, this is due to the technical details of reinforcement learning and its use of verifiable rewards. But partly, it's also because these use cases are not sufficiently prioritized by the companies in their hillclimbing because they don't lead to as much $$$ value. The goldmines are elsewhere, and the focus comes along.
So that brings me to the second group of people, who *both* 1) pay for and use the state of the art frontier agentic models (OpenAI Codex / Claude Code) and 2) do so professionally in technical domains like programming, math and research. This group of people is subject to the highest amount of "AI Psychosis" because the recent improvements in these domains as of this year have been nothing short of staggering. When you hand a computer terminal to one of these models, you can now watch them melt programming problems that you'd normally expect to take days/weeks of work. It's this second group of people that assigns a much greater gravity to the capabilities, their slope, and various cyber-related repercussions.
TLDR the people in these two groups are speaking past each other. It really is simultaneously the case that OpenAI's free and I think slightly orphaned (?) "Advanced Voice Mode" will fumble the dumbest questions in your Instagram's reels and *at the same time*, OpenAI's highest-tier and paid Codex model will go off for 1 hour to coherently restructure an entire code base, or find and exploit vulnerabilities in computer systems. This part really works and has made dramatic strides because 2 properties: 1) these domains offer explicit reward functions that are verifiable meaning they are easily amenable to reinforcement learning training (e.g. unit tests passed yes or no, in contrast to writing, which is much harder to explicitly judge), but also 2) they are a lot more valuable in b2b settings, meaning that the biggest fraction of the team is focused on improving them. So here we are.
Ben Fleming retweeted
This guy re-created all the adobe apps with opus-5.5, ported them to rust, and opensourced them
something really cool is happening
Readers added context they thought people might want to know
These projects are partial open-source Rust reimplementations of Adobe apps (built rapidly with AI agents), not complete recreations; their READMEs and roadmaps document many missing features and features still in development.
github.com/storytold/phot…
github.com/storytold/vect…
github.com/storytold/vect…
github.com/storytold/film…
github.com/storytold/film…
github.com/storytold/ligh…
getartcraft.com
Ben Fleming retweeted
You can get 95% of the results you want by calmly putting one foot in front of the other. One former Navy SEAL friend texted me a principle used in their training: “Slow is smooth. Smooth is fast.”
Ben Fleming retweeted
Huge upgrade for local AI: Apple introduces M6 and M5 Ultra for a "big leap in performance and AI compute." Capable of running even quantized 70B-class models locally
The new M6 model offers (tl;dr):
- Up to 4.8x faster LLM prompt processing than M4
- 42% more memory bandwidth at 170GB/s
- 12-core CPU and GPU
- Dual 16-core Neural Engine
- Up to 32GB unified memory
The M5 Pro version goes up to 64GB RAM and 307GB/s, making it capable of running even quantized 70B-class models locally.
The catch: 32GB limits the regular M6 to smaller models. It should be excellent for 7B–14B models and usable for quantized 30B models, but serious local AI users should choose the 64GB M5 Pro. Anyways, really really cool release!!
Ben Fleming retweeted
There is no AI race without an energy race underneath it.
UK electricity: $0.35/kWh. US: $0.16. China: $0.08.
A frontier GPU cluster is a machine for converting electricity into intelligence. You can fund all the GPUs you like through the Sovereign AI Fund, but if the electrons cost 4x what your competitors pay, the economics never close.
Last year we paid £1.5bn to switch off wind farms and burn gas instead, because the cables couldn’t carry the power south. Connection queues run to 15 years.
None of this is physics. It’s a build problem.
Wrote up what we’re doing about it at Fuse
sotaletters.substack.com/p/t…
Ben Fleming retweeted
Most information is time-consuming, negative, irrelevant to your goals, and outside of your influence.
Ben Fleming retweeted
Chips and drones to be at heart of Burnham’s push to ‘reindustrialise’, AI minister says ft.trib.al/xk4kYMq
Ben Fleming retweeted
The UK is producing unicorns in robotics, chips, frontier AI, and defence.
Unthinkable 5 years ago.
LETS GO
Today, we're announcing that we’ve raised $152 million in our Series A at a $1.35 billion post-money valuation, becoming Europe's first pure-play humanoid robotics unicorn just two years after founding. This funding brings the total raised to date to $270 million.
The round was led by @PrimeMoversLab, with participation from @SchaefflerGroup, @BoschGlobal, Fubon Financial Holding Venture Capital, and Aglaé Ventures.
The capital will support development and launch of the next-generation humanoid robot platform, the roll-out of Beta version in Q4 2026, start of mass manufacturing of wheel-based robots and the continued development of KinetIQ, our AI brain.
Humanoid's valuation reflects rapid execution and the technology's shift from research to commercial deployment, demonstrating that Europe can build and scale globally competitive Physical AI companies.
Read more here: thehumanoid.ai/humanoid-rais…
Ben Fleming retweeted
x is low ego arxiv
hello there the jacobian conjecture is false thanx to my close friend akhil for asking about it and my other close friend fable for working during the world cup final
((1+xy)^3 z + y^2 (1+xy) (4+3xy), y + 3 x (1+xy)^2 z + 3 x y^2 (4+3xy), 2 x - 3 x^2 y - x^3 z): \C^3\to \C^3, has jacobian determinant -2, and sends (0, 0, -1/4), (1, -3/2, 13/2), and (-1, 3/2, 13/2) to (-1/4, 0, 0)
Ben Fleming retweeted
NEW blog post is up!
"Better Than Free: How To Differentiate In The Age Of AI"
Kevin Kelly (@kevin2kelly) is Senior Maverick at WIRED, which he co-founded in 1993, and he is the author of one of my favorite essays on making a living as a creator: “1,000 True Fans.”
But if 1,000 True Fans tells you who your customers might be, how do you figure out what to sell them, especially in the age of AI?
I recently published “Has AI Already Killed How-To Nonfiction?“, and reader Andy Vaughn recommended another Kevin Kelly essay in the comments: “Better Than Free.”
“Better Than Free” answers the what question, and I was shocked I hadn’t seen it before, as it first came out in 2008.
As luck would have it, I know Kevin, so I asked him if he’d like to update or revise the piece. He did so and kindly offered to let me post it here.
When perfect copies of your work cost nothing to make and distribute, what can you actually sell? In 2008, the copy machine was the internet. In 2026, it’s AI, which can produce not just copies but competent variations—of words, images, music, code, and advice—in seconds.
Fortunately, there is still terrain left for humans.
Continue reading at the link below 👇
Ben Fleming retweeted
Introducing Kimi K3: Open Frontier Intelligence
🔹 2.8 Trillion Parameters, 1 Million Context, Native Multimodal
🔹 Kimi Delta Attention enables up to 6.3x faster decoding in million-token contexts
🔹 Attention Residuals deliver ~25% higher training efficiency at <2% additional cost
🔹 Built for long-horizon agentic coding and self-evolving workflows
Kimi K3 is now live on on Kimi.com, Kimi Work, Kimi Code, and the Kimi API.
Open Weights by July 27, 2026.
🔗 API: platform.kimi.ai
🔗 Tech blog: kimi.com/blog/kimi-k3
Ben Fleming retweeted
The US just banned its best AI model, Fable 5. Kimi K2.7, a free Chinese one is already right behind it.
The open-weight model Kimi K2.7's just scored second on ErdosBench. Better than GPT-5.5 xhigh.
ErdosBench evaluates models on hand-crafted proof problems that aren't in any training data, so it's measuring real reasoning.
The single best model on the planet is locked away, while a free, open-weight Chinese one is sitting right behind it and ahead of everything else out of the US.
> LLMs are commoditizing
> Startups like Strawberry will soon be post-training and hosting our own open-weight model forks
> US keeps torching its trade relationships and undermining its AI companies
> China moves towards becoming the world's de facto economic superpower
I really hope Fable becomes globally accessible again, it's a great model and the world should be able to use it.
Blocking your own best model while open Chinese ones catch up makes no sense.
Ben Fleming retweeted
Excited to welcome Andrej to the Pretraining team! He'll be building a team focused on using Claude to accelerate pretraining research itself. I can’t think of anyone better suited to do it — looking forward to what we build together!
BREAKING: Sequoia and Lightspeed co-lead Europe's largest seed funding round with $1.1B at $5.1B post for ex-Deepmind David Silver's Ineffable Intelligence.
David is committing to giving away 100% of the money he makes from his Ineffable equity via Founders Pledge - the biggest pledge in their history and it is likely to amount to multiple billions.
Ben Fleming retweeted
A secretive 4-year old AI company incorporated in the UK has just raised $500m at a $4bn valuation.
The startup - Recursive Superintelligence, is building a new kind of AI that teaches itself.
One of the founders, Tim Rocktäschel (@_rockt), is a professor of AI at University College London and was, very recently, a principal scientist at @GoogleDeepMind.
The rest of the team is also STACKED and includes researchers from Meta, OpenAI and Google.
The FT (linked below) has reported that the company has raised $500m at a $4bn valuation led by @GVteam.
The round was so oversubscribed that it could end up raising $1bn.
According to the FT:
"Recursive has not yet officially announced its existence, but a filing at the UK’s Companies House register shows that it was incorporated in London at the end of last year"
Amazing news if the company is actually HQ'ed in London.
The city is having a huge tech resurgence and we LOVE TO SEE IT
Amazing stuff as always from the FT team @tim @IvanLevingston @GeorgeNHammond
Ben Fleming retweeted
UK AI startups are flying
In Q1 they raised $5.8bn up 176% YoY.
Nscale, Wayve, Synthesia, ElevenLabs all raised big rounds.
And tomorrow Joséphine Kant, @Jameswise and the rest of the @UKSovereignAI team is officially launching its £500m fund to invest in British AI companies.
It's great to see the UK taking AI seriously.
LETS GO 🇬🇧
Data from @yoramdw and the @dealroomco team
France is transitioning government desktops to Linux, with each ministry required to formalize its implementation plan by autumn 2026.
linuxiac.com/france-launches…
#Linux #OpenSource






























