Dagens Vibes — 6. juli 2026

Meituan LongCat@Meituan_LongCat

🐱 LongCat-2.0 is now fully open-source — MIT licensed, no restrictions. Since our launch a few days ago, the response from the community has been incredible. Thank you for all the feedback, discussions, and interest. Today, we’re releasing the model weights and inference code to everyone. ◆ 1.6T MoE · ~48B active · 1M token context ◆ Agent-native: Integrates directly with Claude Code, OpenClaw, and Hermes Agent ◆ Deployment: Support both GPU and NPU platforms— verified on large-scale domestic clusters 📑 Tech Blog: https://t.co/W9EYWDW3Y9 🤗 HuggingFace: https://t.co/EP2szwIhu2 💻 GitHub: https://t.co/qJzXMQcC2z 🪄 ModelScope: https://t.co/nKt5Wh3m8Y 👇 Inference Code GPU: https://t.co/Tq8QKvneH6 NPU: https://t.co/l96ebhFxN6

Lachlan Donald@lox

Built https://t.co/FkfBhMA9sP for @AmpCode on the weekend (ported from pi's) to make use of Fable as oracle. You give the agent a benchmark and it loops: edit, measure, keep or revert, annotating dead ends, with a nice TUI/web dash. Pointed it at sporevm's already heavily optimized cold TTI and it went 77ms → 40ms in 29 runs, then asked the oracle to check its own tradeoffs. 📉

Anthony Kroeger@kr0der

Cognition's new Devin Fusion harness looks like it's gonna be really good instead of swapping models mid-chat (which busts the cache and is expensive), it runs a smaller "sidekick" model in parallel with the main model, and the main model delegates tasks to it + reviews the work. also, if a task suddenly becomes complex, it brings work back to the main model. this seems a lot better than the current way of doing auto-routing which determines which model to use for an entire prompt based on assumed complexity of the task. imagine your task seems simple so it routes to a cheap model but it ends up being complex - you either have to restart your entire chat or swap models mid-way which can be expensive due to breaking the cache

Philo Groves@PhiloGroves

GLM 5.1 + a Tencent cyber harness is now #4 (84%) on CyberGym. GLM 5.1 originally benched at 68% with Claude Code as a harness. Tencent claims the jump is due to a collection of cyber skills and subagent orchestration. https://t.co/vM9BMo608C https://t.co/kkDKrYn9fh

Matt Pocock@mattpocockuk

This is a PRD based on a multi-day /wayfinder session Look how detailed it is Look how every assertion is linked back to the session where it was decided Secondary source -> Primary source Beautiful https://t.co/AMVKBNJq6X

Matt Pocock@mattpocockuk

The "should you read code" debate is dumb because the real decision isn't binary, it's a scale: 1. Reading every line of every diff 2. Scanning every diff, reviewing important lines 3. Ignoring diffs but understanding the 'why' of every PR 4. Spot checking PR's instead of reading every one 5. Ignoring PR's, but doing regular spot checks on the codebase 6. Ignoring the code, but spot checking agent traces to help improve the system 7. Ignoring both the code and the system, let models handle everything Where are you on the scale?

David Fowler@davidfowl

I see a lot of influencers talking about “not reading the code” and “building software factories” but I don’t see much discourse about what software engineering teams of the future look like? Software is a multilayer game as much as everyone will try to convince you that all you need is a team of agents…

Dillon Mulroy@dillon_mulroy

yall im ngl its way harder to get joy and satisfaction out of building with ai than it was before constantly straddling burn out, being far less immersed in hard problems, constant context switching i’m tired of the uncertainty of where this is going and how to do it well

Peter Yang@petergyang

"Codex is the only way I can keep track of everything." From Rohan (Codex PM): "Whenever I’m in meetings, or literally anytime I don’t understand something, I just have Codex go and gather all the context. We're super lucky because we're building a tool for ourselves." 📌 Full episode: https://t.co/jPa0m1bmCE