It's official. I've joined the @ethereumfndn AI team to make Ethereum the trust layer of the agentic economy. The AI economy is just getting started, and Ethereum is the perfect place to coordinate it - excited to push this forward. Send a dm if you're building cool stuff.

Today is my last day at @Alchemy, after more than four years, and I have nothing but gratitude for this adventure. I need to thank the people who made it one of the best chapters of my life 👇 None of this would have happened without the GOATs @VittoStack and @thatguyintech. It all started with the unforgettable Road to Web3 bootcamp and the first cohort of Alchemy Ambassadors. They gave me a place in this community, and then asked me to join the team. Then came @BeingDanNolan, @CodesMcCabe and @punk6068, three incredibly talented and kind people behind Alchemy University, a project that opened the door for so many students and new builders. I was lucky to collaborate with them and support the students along the way. Big shoutout to the marketing team too, some seriously talented and committed people I learned so much from: @itsNikkiLOL, @swampnet, @MaxCrawford8, @cryptobytch, @uttam_singhk and many others. Thank you @nikil and @thejoelau for building Alchemy, a project I'll always love, and for giving me the chance to be part of it. And a very special thank you to @angelinarusse, for always being so kind and making me feel like part of the family. But most of all, the Alchemy community 🙌 The kindest, most talented and most human group of people I've ever met. You were the reason I showed up every day, to every session, full of energy and the joy of doing it together. @talondragon000, @bartomolina, @LTebbs2, @AnshMinchekar, @naruto11eth, @RocioGonzalezT9, @spaceagente, @stevibe, @artlu99, @yikesawjeez, @FrancescoCiull4, @KillianeMenand, @BMbotsan, @chai_really, @tusharlog, @SummerMaxis, @0xeik, @SahilAujla15, @davide_petti, @emmanuelkiranrand and many many more! 🫶 Thank you from the bottom of my heart ❤️ PLW3! (My DMs are always open, let's keep in touch!)

I want to share my hot take on the TypeSafe Jev model that was announced yesterday. Comparing it to general LLMs is misleading because the model cannot generate free text. It's more similar to a smart tool. Based on the (little) public information it seems more similar to an old school classifier. It has some cool features, it leverages the flexibility and parallelism of transformers and can take free text input, and is probably easy to adapt to specific use cases that require structured output (you can get a tailor classification-style output quickly and cheaply). These tools are interesting in applied environments: - Business operations - Data infrastructure - Robotics processes as they can generate hard surfaces at lower cost with more security guarantees. But IMO not a game changer in terms of Frontier AI as the casual twitter observer would think.

bringing this line back from Feb that is still proving to be true: "The dividing line has never been skill. It has always been agency - the capacity to perceive your situation accurately, make decisions under uncertainty, and act on those decisions before you're forced to." -- @VittoStack

Introducing GLM-5.3: Built to Code. Ready for Cyber Defense. - Top-tier coding and agentic capabilities, achieved through post-training on the 743B base model - A major leap in cybersecurity, setting a new standard among open models Tech Blog: z.ai/blog/glm-5.3

Qwen3.8-27B Abliterated is now available for Apple Silicon with MLX 🔓🍎 Separate 4, 6, 8-bit + BF16 releases. All four scored 0/100 explicit refusals on harmful prompts and benign controls, plus 12/12 quality, 8/8 tools and verified 4K retrieval. huggingface.co/PocketAiHub/Q…

some not great news: i'm no longer at cyfrin due to engineering layoffs. i am open to full-time/part-time/contract full stack or forward deployed engineering roles. if you're hiring OR can share this, i'd appreciated it 🫶 leading @CyfrinUpdraft for 2.5+ years was a dream. i'll cherish the time there and the people i worked alongside. seeing products grow because of the people using them is why i love this industry. more on my work below ↓

guys you do know you can just disable thinking, and instead give it a "deep_think" tool, and it will call it with internal CoT reasoning format right? gl fixing that

What are the best models you can run on your @NVIDIAAI DGX Spark? ✨ Aug 2026 Edition 1× DGX Spark • DeepSeek v4 Flash - 1M ctx, 26 tok/s - recommended! • ⁠Qwen 3.6 35b NVFP4 - 256k ctx, 81 tok/s • ⁠Qwen 3.6 27b NVFP4 - 256k ctx, 33 tok/s ⁠Qwen 3.8 27b - should be released soon and this could change my recommendation! 2× DGX Sparks ← sweet spot! • DeepSeek v4 Flash 0731 - 1M ctx, 82 tok/s • Inkling-Small - 1M context, Full Omni, 33 tok/s • MiMo-V2.5 - 1M ctx, Full Omni, 31 tok/s • Step-3.7-Flash — 256K ctx, 30 tok/s 3× DGX Sparks • GLM-5.2 with Vision - 348k context, 25 tok/s - still the best intelligence you can run locally if you have 3 sparks. • DeepSeek v4 Flash on 2 units + smaller models on the 3rd spark for images, ComfyUI and other things. This setup gives you DeepSeek v4 Flash speeds for coding, plus strong agentic workflows and image support from smaller models. 4× DGX Sparks • GLM 5.2 NVFP4 across all 4 units - still think this is the way to go if you have 4 units! • The alternative is running DeepSeek v4 Flash on 2 units and any other 2x setup on the other units. The choice is your. Links and repos below 👇