Stefan

@ethPandaOps at Ethereum Foundation, prev @etherchain_org

[turbopuffer @turbopuffer]

now in beta: native embeddings in tpuf

embedding is the most painful part of puffing. we want to make it easy

you can now convert chunks to vectors as you read and write to turbopuffer, without extra calls to an embedding model provider API

docs: https://t.co/iC5ZJK5Xkh

[Dankrad Feist @dankrad]

Nice! Ethereum devs getting more ambitious.

300M gas limit for Glamsterdam?

[Stefan @stefan_star]

@EFprotocol @abcoathup 🫡

[Stefan @stefan_star]

@wolovim @ethereumfndn sorry to hear :/, appreciate all your amazing work 🫡

[Stefan @stefan_star]

amazing work

[marc | wolovim.eth @wolovim]

unfortunately, i was among those laid off from the @ethereumfndn in the last round.

over 8+ years, it was a hell of a ride and a genuine joy to work with all my colleagues across three teams (mist browser, python tooling, protocol support) while at the EF; their blend of humility and ambition will continue to inspire my approach to work.

immensely grateful to have landed with the protocol support team last year, to whom i pitched my little side project, Forkcast. it felt like a real spark from the start, but Forkcast has grown into something i'm deeply proud of: a tool that is relied on by Ethereum's stewards to understand and react to their complex world.

i wake up and go to bed thinking about the nuanced social challenges that Forkcast attempts to tame. the work is meaningful and demands the wide range of technical and non-technical skills i've developed over a 12-year software career. its difficult to imagine stepping away. fortunately, i don't have to just yet.

starting with the next three months, i'll be an independent contributor to Forkcast and the ethereum/pm repo, with support from the EF. during that period, i'll also explore what comes next for me, possibly including spinning out Forkcast. the platform makes good sense to me as an independent observer of Ethereum core development and i'd like to hear from individuals and orgs in the ecosystem that want to support that work. gratitude to those that have already reached out to start conversations.

DMs open if you want to chat about the future of Forkcast or ways we might work together.

[Marek Moraczyński @M25Marek]

Testing @AnthropicAI's new Fable 5 model.

My prompt is "Implement Ethereum Strawmap".

[Stefan @stefan_star]

Sonnet-5 does not impress and its just as slow as opus.

[paji.eth @paji_a]

直近のAI情勢…

[James Rosen-Birch ⚖️🕊️ @provisionalidea]

ngl it’s kinda wild that China is the land of hypercapitalist competition fueled by open-weight models anyone can use, and America is the land where the executive branch of government must personally approve you to have the privilege of giving a private company your money

[banteg @banteg]

the best part is you can download parquets and have all the verified contracts locally, open verification is the future

[Stefan @stefan_star]

@Hikari_07_jp @Tono_Ken3 nice ive pused it to 130 decoding tokens/s on the same setup MTP on 2. Mostly did some autoresearch and created a SM120 syllabus.

[Stefan @stefan_star]

@MiaAI_lab its such a great model for its size

[Stefan @stefan_star]

Not even Fable can solve uptime

[jack friks @jackfriks]

claude is down with a major outage for everyone

except for the government

[CoinMarketCap @CoinMarketCap]

LATEST: ⚡️ Five former Ethereum Foundation researchers have launched Ethlabs, an independent nonprofit research hub backed by SharpLink, Bitmine, and Consensys CEO Joe Lubin.

[Stefan @stefan_star]

Does it have airconditioning?

[Tim @TimurNegru]

A château with its own chapel, stables, a tennis court, and a private dock on a canal.

I don't think it's missing anything tbh.

640m² (6,889 sq ft), 8 bedrooms, 6 bathrooms. Built in 1850, fully restored between 2020 and 2025. Most châteaux on the market still need that work done, but not this one.

3.15 hectares (7.79 acres) of parkland, five-stall stables, a private chapel from 1867, a natural spring and direct mooring on the Canal des Vosges.

€1.2M ($1.37M).

The only downside I see is that this is rural eastern France. Accessed through the forest and with no neighbours. The nearest village is a few minutes away, but Épinal or Vittel, the closest towns, are under an hour's drive.

Whoever buys it will get themselves a truly exceptional property and for a good price too.

[Vite ⚡ @vite_js]

🚀 Vite 8.1 is out

⚡ Experimental Full Bundle Mode: ~15x faster dev server startup & ~10x faster full reloads on large apps
🧩 Chunk Import Map to keep chunks stable & improve caching
🦀 WASM ESM integration
💡 Lightning CSS on the road to default
👀 and more!

Blog post below.

[Franz Branntwein™ 🍀 @FranzBranntwe10]

Meteorologen warnen in Deutschland vor 40°C.

Nach Steuern und Abgaben kommen da etwa 20°C beim Bürger an.

[Stefan @stefan_star]

updated 🫡

[sproul @sproulM_]

Up to 40% of my LH peers haven't updated to 8.1.3, if that's you, please update to 8.2.0.

In the new LLM era, there are vulns coming in like crazy and we need everyone to keep their nodes up to date!

[Lorenzo Valente @LorenzoARK]

On the Ethereum funding problem:

the ethereum funding debate is getting spicy and the best counter i've seen argues the real problem isn't funding, it's conviction. The linux kernel analogy: nobody pays kernel devs, yet they never lack contributors, because contributing to linux is so obviously valuable it converts directly into career capital. same with early ethereum, people built for free in 2016-2018 because the perceived value of participation made compensation irrelevant.

I think this argument has a lot of holes:

the linux comparison falls apart when you look at who actually maintains the kernel: red hat, google, intel, microsoft, all paying engineers full-time because it serves their commercial interests. ethereum client teams don't have that backstop. the "conviction converts to career capital" argument works great for devs shipping dapps. it does not work for consensus engineers doing maintenance on nimbus or teku that literally nobody outside the protocol layer will ever see on a resume.

the 2016-2018 nostalgia is also doing a lot of heavy lifting here. that era had a massive ICO bubble pumping perceived upside, ETH going from $10 to $1,400, and a much simpler codebase. you could contribute meaningfully in your spare time. today's ethereum needs deep specialization in distributed systems, cryptography, and formal verification. that's not getting done nights and weekends by passionate volunteers.
and the causality runs both ways. declining conviction causes underfunding, sure. but underfunding also causes declining conviction. when core dev slows, upgrades take longer, the protocol feels stagnant, and conviction drops with it. you don't let the plumbing rot while you sort out your narrative.

these protocols are far from their end state and what worked in the early days probably doesn't work now. ethereum has changed enormously, the complexity, the stakes, the competition. i don't think there's a clean answer here, but finding a more sustainable way to fund the ecosystem isn't optional. it's overdue.

[Stefan @stefan_star]

fml

[Sakana AI @SakanaAILabs]

Introducing Sakana Fugu: A full multi-agent orchestration system accessible via a single model API.

Our ‘Fugu Ultra’ model matches the performance of Fable and Mythos, delivering frontier capability without the risk of export controls.

Try it: https://t.co/hhO6qTawgb 🐡