I made a visualization of EIP-8288 aka "recursive mempool aggregation," my new favorite EIP of all time because it perfectly encapsulates the thesis of the "cryptographic world computer" -- a system whose scalability *increases*, not decreases, with the level of decentralization. In a previous tweet I explained how EIP-8288 achieved this seemingly impossible feat, but one gif is worth a thousand words. Essentially, EIP-8288 leverages Ethereum's decentralized mempool as a parallel ZK proving cluster. Clients send signatures and proofs to the mempool, which over multiple rounds progressively aggregate them into fewer and fewer proofs, until the block builder finally aggregates them into one SINGLE proof. This design accomplishes five things at once: - The more nodes we add to the decentralized mempool, the more signatures/proofs Ethereum can process as a whole. This is obvious if you look at the animation and imagine more nodes in the mempool ring, which would "fan out" to more signatures/proofs. - By aggregating over multiple rounds (aka "recursive aggregation"), the bandwidth consumed by proofs remains constant for each node even as the network processes more proofs. As a result, the network stays decentralized even as capacity increases. - By aggregating proofs before one final aggregation by the builder, as opposed to requiring the builder to aggregate everything, the network can process more proofs without increasing hardware requirements for the builder. - By aggregating everything to a single block-level proof, validators can verify thousands of signatures/proofs by checking just one aggregated proof. Hardware requirements for validators can therefore stay low, keeping the network decentralized. - By aggregating not just native signatures but also STARKs, Ethereum can support any signature scheme by wrapping them in STARKs. So once again, this is how Ethereum will: - scale to process tens of thousands of PQ and private transactions per second - charge a tiny amount of gas to validate each txn, even though the original signature/proof may be huge (up to hundreds of KBs in size) - all while decentralizing even MORE. You are not bullish enough.

It's Official

🚀 Introducing DeepInfer, our in-house inference framework, now fully open source! Powering DeepSeek-V4.1-Flash at scale with high throughput and low latency. We’re sharing our production stack to help the community lower compute costs, reduce energy use, and make AI more affordable. 🔗 github.com/deepseek-ai/DeepI… 1/4

Over the past few months, the @Nethermind Client team has been focusing on almost everything BUT performance - authoring, supporting and reviewing EIPs for Glamsterdam, championing Hegota headliner - Frame Transactions, researching state bloat, and much more. That's not changing - we are still being heavily involved in Ethereum roadmap. But AI has unlocked a whole new level of productivity for us, and we're now pushing hard on several fronts: 1. zk guest program - coming soon to @eth_proofs. Very promising, very performant. 2. RPC refresh - Geth compatibility, major performance improvements, and more. Few projects already got a feeling of what we are shipping - and it is looking very good. 3. Client hardening with AI - Node MCP for users and our custom CoreMCP for developers, helping us ship faster and build a much more stable client. 4. Innovative node optimizations and advanced plugins - unlocking capabilities that could be extremely useful for many projects across the ecosystem, including support for new chains. And that's just the beginning. As a "side effect," we've also significantly improved the node's performance at the tip of the chain - before even getting into the Glamsterdam hard fork, which should unlock a whole new level of throughput. Nethermind is the fastest #Ethereum execution client, and the only one pushing beyond 1.5 GGas/s in p50 block throughput at the tip of the chain. With the next fork, we expect that ceiling to go even higher - based on Sepolia data, potentially around 3x higher. And we're definitely not done. We want to change the rules of the game. The AI-powered tools we've built in-house are giving us enormous leverage, and we're confident this advantage will only grow over time. Performance is good. 🚀

RT @tianyi: DeepSeek just open sourced its high-performance, large-scale inference framework DeepInfer in Apache-2.0 license. https://t.co…

JUST IN: JPMorgan announces it is bringing JPM Coin to Ethereum, allowing institutional clients to transfer U.S. dollar deposits in seconds, 24/7.

BREAKING: 🇺🇸 $5 trillion JP Morgan is expanding its JPM Coin to Ethereum. First, banks ignored crypto and now they are going all in.

JUST IN: JPMorgan has launched its stablecoin 'JPM Coin' on Ethereum

🔥 JUST IN: JPMorgan is bringing JPM Coin to Ethereum for institutional clients, per Bloomberg.

JUST IN: JPMorgan expands crypto deposit token JPM Coin to Ethereum, allowing institutional clients to transfer money 24/7.

The liquidity is on Ethereum.

摩根大通通过其区块链业务部门 Kinexys正式将美元存款代币JPM Coin扩展到以太坊主网，支持机构客户之间在 EVM生态内的点对点转账。关于为什么选择以太坊？官方的说法是“Ethereum's established depth of onchain liquidity and tokenized assets makes it a natural next public blockchain for JPM Coin”（以太坊已有的链上流动性深度和代币化资产生态，使其成为自然的下一个公共区块链）。 核心就是流动性、代币化资产生态、机构采用度、技术兼容性和安全性上的综合优势，也能与银行现有产品形成闭环。

News: JPM Coin (JPMD), @jpmorgan's USD deposit token, is now live on @ethereum. JPMD extends from Base to Ethereum L1, giving institutional clients native onchain cash that sits alongside the assets, collateral and counterparties they transact with already. J.P. Morgan has now added a cash leg to the products it already live on Ethereum mainnet, with tokenized money market funds MONY and JLTXX both already on the network. Ethereum is the home of onchain finance.

Ultimately, precompiles and EVM are both the wrong place to put most cryptography support into Ethereum. The reason is that precompiles and EVM both operate on data that has to be included into the ethereum chain somehow, which realistically means included in a block as calldata. That space is limited - even with PeerDAS and blocks-in-blobs, there's a practical cap of ~8 MB/s before it starts getting uncomfortable. Remember: it's not a blockchain. It's a cryptographic world computer. Any data/computation that does not need to be in the "main thread" of data/execution should not be put into the main thread. Instead, the correct solution is... EIP-8288 maximalism. If you have any heavy computation that's self-contained, compute it and STARK it client-side (this is quickly getting faster and faster, even on phones), and push the STARK through the EIP-8288 recursive aggregation mechanism. The chain sees the fact that the computation was done and what the result is; it never needs to see the actual data and computation. This way everyone can use whatever cryptography they want, the Ethereum chain can be lean, and we never need to add more precompiles again.

ETH是真有用， 作为全球计算机， 大家都需要用， RWA万物上链以后， 更是蓬勃发展， 所以不需要搞什么虚的精神。 并且ETH又有手续费烧币这种价值捕获， 虽然现在抢蓝海市场、 薄利多销，没收多少钱， 但以后快速增长期结束， 应用量顶到TPS硬件上限， 链上繁忙，开始收钱时， 就超级爽了。 BTC是真没用， 所以必须搞点虚的比特币精神， 维持 「有价值是因为屯币」 「屯币是因为有价值」 这种循环论证。 如果连这点虚的比特币精神都没有了， 那就真一文不值了 ： ）

You dont need to upgrade your memory. 64GB is just fine to run Q4

批评讽刺 DeepSeek"蒸馏"的现象，是人类科技史上最大的黑色幽默。 DeepSeek 前年搞开源周，一周五天，FlashMLA、DeepEP、DeepGEMM、DualPipe、3FS，训练和推理的底层代码一天放一个；R1 的论文把训练方法写得清清楚楚，上了《自然》封面，是第一个过同行评审的主流大模型；去年最后一天梁文锋亲自署名发 mHC 新架构论文，后来直接用在 V4 上。整个模型权重扔网上，随便你微调、继续训练、拆了魔改，从来没见哪家开源公司出来骂过。 天天跳脚骂街的都是什么货色？OpenAI 自己偷纽约时报数据被告到现在，转头给美国国会写备忘录告 DeepSeek 的状。Anthropic 更绝，去年刚因为从盗版网站下了几十万本书，掏了 15 亿美元跟作者和解，今年就点名 DeepSeek蒸馏他数据。 闭源模型靠“蒸馏”全世界盗版数据才有的先发优势，如今卖 API（还肆意降智封号缩水）大赚黑心钱，现在却满世界贼喊抓贼。道德水平略低于旧社会黑市销赃的二道贩子。

看了一天的 DeepSeek 新模型有感 DeepSeek 开源了不少论文 也帮助不少厂商降本增效 为什么现在只听到蒸馏的指责 却很少看到感谢？

笑死，老外实在看不下去 DeepSeek 的模型命名了，主动请缨帮忙起名字。 国外 Claude 和 GPT 的起名风格： Claude 是文艺青年风，吟诗作赋。 Opus（巨著） Sonnet（十四行诗） Haiku（俳句） GPT 是星际旅人风，探索宇宙。 Astra（星辰） Sol（太阳） Terra（大地） Luna（月亮） DeepSeek： DeepSeek-R1 DeepSeek-V3.2-Speciale DeepSeek-V4-Pro DeepSeek-V4.1-Flash-1017 我愿称之为国内程序员风，朴实无华，严谨务实，写好版本号，标好日期。

Deepseek v4.1 flash 1017 猫榜编程测试出分了，提升可观 实打实的超过了kimi k3 和 Glm 5.3 接近gemini 4 argon 的水平。。 再说一次，deepseek 模型几乎从来不刷分。。。 因此v4.1 flash 1017 的能力提升是很大的。。 个人认为从综合实力上来说，和 GPT 6.1 Sol应该是接近的