Yukon Weekly Winners, Sept 11 to Sept 17. $10,000 in prizes. Last week Meganpark980320 won a raffle spot with 698 points. This week they are number one with 14,424. Points come from landing verified improvements on live challenges. Everything that lands stays in the open for the next person to build on. This week's round is open.

New on Yukon: the Quantum Safe Bitcoin Challenge, with @StarkWareLtd. On August 26 the first quantum-safe Bitcoin transaction landed on mainnet. It is locked with a hash instead of the usual signature math, and it works under Bitcoin's rules today. No soft fork. That transaction cost around $320 in GPU time, and about $280 of it was one step: searching for the right combination of signature pushes to omit. That step is the challenge. Score is verified candidates per second on one GPU, every hit rechecked on CPU, judge-owned clock, fresh problems on ranked runs. Speed up the search and almost the entire cost comes down with it.

World’s first massive multiplayer research. 500+ improvements made by 100s of humans and their agents over many months to figure out how to build Quantum circuits that can break https and Bitcoin. This was the challenge. We started at 0.7x relative efficiency to Google and the system improved it to 2.63x! Given the OpenAI-Anthropic tussle on Navier-Stokes credit, this kind of system would have helped to assign credit and encourage collaboration. AI-native scientific institutions are coming. They are sorely needed now. Go checkout

The tiling patterns being generated in the Heesch challenge @yukonresearch are so beautiful. I took the promoted submissions from the competition till now and made an infographic around what the tilings took like to visually understand them. There is also active github discussions going on at shepherded by @nasqret, with the goal of coming up with a piece that would allow for being able to create a tight 5-layer without any gap. The constraint is that piece has to be composed of squares, hexagons or triangles. Bartosz has said that finding such a piece will be a real mathematical discovery. You can see in the infographic where the gaps are in existing promoted submissions, and it seems we are very close to 5.

It's important for us to talk about the safety risks of AI swarms. It's also important for us to talk about the positive things they can unlock. Today, a paper was released on arXiv chronicling what researchers believe is the first large scale swarm deployed to push the frontier of quantum computing. Earlier this year, we at @YukonResearch launched opening a frontier quantum computing problem to 100+ humans and AI agents. Participants collectively pushed the best known result to >50% below Google Quantum AI’s reported frontier. The paper released today by @jeiyilong is the first formal write-up of what came out of that swarm. There’s a lot we still need to understand about swarms. But it feels like this is an early glimpse of how important they will be for the future of science.

If you can coordinate people in masses flexibly, you can move mountains. We brought together thousands of people from diverse backgrounds (professors, engineers, cryptographers, landscape designer, 13yo high schooler, farmer, marketer) to solve this one problem. All our past work over 4 yrs revolves around that and two words: Open Innovation. We did that for compute (@DarkbloomAI) and capital before. Intelligence was the hardest format to crack and there have been different mechanisms. ECDSA(.)Fail was the first time where it worked at a huge scale. Thousands participated, hundreds got their contributions in. The format was pretty simple: have a hill-climbable goal and verifier that scores your work. AI played is important role in this - I’m the top of the leaderboard in terms of % of contributions yet I don’t know anything about quantum computing, barely about cryptography and high school dropout level math. I just knew how to build good harnesses (pre Astra, Fable etc) - Opus 4.6 is what we used mostly. The result of bringing everyone towards to solve a problem is the most exciting thing in world. We are doing that with Darkbloom and @YukonResearch (evolved and multi challenge scalable version of ECDSA Fail). We started with cryptography, did inference optimizations and we will continue to solve the hardest problems for humanity together. The beauty of living in the era of abundance. - This is also my first time on ArXiv. Thank you @JeiyiLong for writing this paper.

Welcome 👋 We're launching @Alibaba_Qwen 3.8-Flash-Next on BOTH and today! This is our first challenge on We see a lot of overlap between the MLX and CUDA communities, and we want to see how far both can push the same model. So we're putting their progress on the same graph for this one! Just some friendly neighborhood competition 😉 The challenge: make Qwen 3.8 Flash Next run faster on two concrete setups: --- our Qwen port of @antirez's ds4 C/CUDA engine with @UnslothAI's Qwen3.8-Flash-Next-GGUF on 1x NVIDIA DGX Spark. --- our Qwen Swift/Metal engine, built on the MLX and mlx-swift-lm forks, on 1x M5 128 GB MacBook. Both tracks measure one response stream at a time. Your changes must pass the correctness checks, then beat the reference engine on the same machine. The score combines faster prompt processing (prefill) and token generation (decode), weighted 25% and 75% in the geometric mean. We chose Qwen 3.8 Flash Next because it beats Claude Opus 4.6 Max reasoning on 94% of the benchmarks from Qwen's published comparison. That's 15 out of 16, spanning coding, reasoning, instruction following, and vision! Also this model can be ran on a single Spark or a 128 GB MacBook. That's what makes this worth pushing: every inference improvement makes that capability more useful on developer hardware. Our current starting decode rates are around 17–18 tokens/sec on the Spark and 35–36 tokens/sec on the Mac. Let's see how far we can push both! Thank you to @TheDavidTai and @GumbiiDigital for building the entire challenge (it was not easy!), @antirez @ivanfioravanti & many others for their ds4 work (the Gs), and the MLX and Qwen teams for the work we're building on. One caveat: this is a significantly larger model, so local testing will be harder for some smaller Macs. We recommend a DGX Spark or a Mac with more than 128 GB of memory to give yourself room to work.

what a week to publish this paper has been a live demo of what open, multiplayer research collectives can achieve together. everyday people - both experts and amateurs - directing their agents and harnesses to solve one of the world's hardest problems, sharing their progress, building on one another's learnings and going further together than anyone could have gone alone. proud to have been a small part of a group (and vibrant slack channel) of 100+ solvers experimenting on @yukonresearch and sharing in the struggle and the successes of moving the frontier of quantum cryptography. projects like this are the antithesis of openai's wanton harvesting of scientific progress for their own gain and glory. its what open, scientific progress SHOULD look like. we've got a lot of work to do to make participation more permissionless and ensure that credit is more accurately assigned and rewarded, but early wins like this give me confidence that the best days of open research and science are still to come. huge s/o to @bbuddha_xyz @sreeramkannan @soubhikdeb @SahilDewan @gajesh for their tireless work on the platform @drakefjustin for gathering and rallying the community around this challenge and @jieyilong for distilling the work into this paper!

Couldn't be more proud to have my name in a paper that is focused on how WE CAN DO AMAZING THINGS TOGETHER!!! take the doom messaging and throw out, we belong in a future of hope, compassion, integrity and working hard for each other. This is what makes life amazing. This is just the beginning. Incredibly proud to finally be included in some tiny fashion, into this world. Ty @eigenlabs and everyone involved.

Proud to be a coauthor and friend of @jackieleeeth and @jieyilong (CTO @Theta_Network) on — now on @arxiv. Jackie invited me into the @eigenlabs challenge, and we found solutions to shrink secp256k1 point-addition circuits in Shor's algorithm. Across the leaderboard, humans and AI agents cut the qubits×Toffoli score by 86% — more than 50% below Google's published thresholds (different accounting conventions, so a numerical rather than formal comparison). It was so much fun meeting @jieyilong, @nasqret, @gajesh , @drakefjustin and everyone else along the way. Huge thanks to the organizers for the challenge and the fun, and to this amazing community. Coauthors from @ethereumfndn, @StarkWareLtd, @Starknet, @trailofbits, @brevis_zk, @SeiNetwork, @pauli_group, @OctavFi, @sciencevr, Adam Mickiewicz University, Warsaw University of Technology and Stanford's Free Systems Lab. Reviewed before publication by Craig Gidney and Tanuj Khattar of @GoogleQuantumAI. Full paper:

been hiding something from yall for quite a lot of time. this was one of the most extremely fun effort, that few of us from starkware and starknet were part of. well now we are officially core authors, contributors and researchers as part of the paper. adding this on to my linkedin now brb.

The first full paper is out today. It tells the story of how 100+ humans and AI agents produced a circuit scoring 50%+ below Google Quantum AI’s reported result. And the challenge is still open.

In March, Google Quantum AI proved a record breaking quantum circuit existed, while the implementation was not made public. What they released instead was a proof that the circuit exists. And that can be verified: a program that tells you whether a candidate circuit is correct and what it costs. Which left anyone free to build their own and test it against Google's number.

Bitcoin and Ethereum’s quantum threat is getting harder to ignore. In roughly two months, 100+ participants and their AI agents drove down the estimated cost of the arithmetic of a potential quantum attack on elliptic-curve cryptography. Their work is detailed in the new paper, with seven co-authors from StarkWare and the Starknet Foundation. The challenge focused on a calculation that a sufficiently powerful quantum computer would use to recover private keys from exposed public keys. The result: an 86.1% reduction from the challenge’s starting resource score, which combines logical qubits and key computational operations. That score was also over 50% below Google Quantum AI’s reported result for this calculation, although the circuit setups and measurement methods differ. In simple terms, algorithm improvements can reduce the resources needed for an important part of a quantum attack. For StarkWare, this complements our broader work on quantum security in blockchain, from STARK proofs and Starknet’s post-quantum roadmap to our successful quantum-safe Bitcoin transaction on mainnet. We’re helping measure the threat and build the defenses. Congratulations to our seven co-authors across @StarkWareLtd and @StarknetFndn: @odin_free, @8am1am, @franklyteddy, @akashneelesh, @0xLucqs, @robertkp13, and @tarekeleter, alongside the wider community. Read the paper: