Dagens Vibes — 18. oktober 2026

Fra X-feedet

DeepSeek opdaterer Flash — og et eksternt review-tjek ser lovende ud

V4.1-Flash-1017 er ude med åbne MIT-vægte, kortere svar og bedre agentresultater i DeepSeeks egne tests. Mere interessant: BugBunnys test af 100 sårbare PR’er fandt 54 fejl for $7,53 mod forgængerens 44 for $12,79 — ét konkret review-benchmark, ikke et facit på al kode.

DeepSeek@deepseek_ai

🚀 DeepSeek-V4.1-Flash-1017 is here! Stronger benchmark performance, faster task completion, and shorter outputs than the previous version—a Pareto improvement in performance and cost. Now live on our website, app, and API. 🛠️ Optimized for @DeepSeekHarness, Tencent CodeBuddy, WorkBuddy, and more. Upgrade to DeepSeek Harness v0.2.1-rc.1 or later for the best experience with the new model. 1/4

♥ 31,634 · ↻ 3,556 · 💬 1,350 · 🔖 5,912

https://x.com/deepseek_ai/status/2111375640717033777

Z A D D Y@Zaddyzaddy

We gave DeepSeek’s updated V4.1-Flash-1017 the same 100 pull requests with known vulnerabilities and an hour per PR to find them in the @BugBunny_ai VulnPR-100 benchmark. It found 54 for just $7.53, making it the best-performing open-weight model on our leaderboard. That matches Grok 4.6’s 54 findings at roughly 16× lower review cost. It also finishes only three findings behind Gemini 4 Argon while costing roughly 22× less. The previous DeepSeek V4.1 Flash found 44 for $12.79. The update catches 10 more vulnerabilities while costing 41% less, bringing the cost down to just $0.14 per vulnerability found. Better recall, lower cost, and the strongest open-weight result we’ve measured. Seriously impressive work from the @deepseek_ai team.

♥ 808 · ↻ 58 · 💬 43 · 🔖 170

https://x.com/Zaddyzaddy/status/2111452091164000261

OpenCode Console ramt: modelnøgler på afveje

OpenCode bekræfter et brud hos 1,4 % af Console-brugerne med eksterne providers tilknyttet; de berørte er kontaktet. Max Weinbach rapporterer tømte API-balancer — det er ikke en melding om kompromis af alle lokale OpenCode-installationer.

OpenCode@opencode

We had a security breach last night related to a large scale distillation attack that impacted 1.4% of OpenCode Console users We have notified them all this morning via email. If you did not get an email you were not impacted, it was a subset of users who had external providers linked The incident is ongoing so we cannot share more details at this point but we are working with Anthropic, OpenAI, and Google to remediate

♥ 8,127 · ↻ 1,075 · 💬 468 · 🔖 1,267

https://x.com/opencode/status/2111519637049319818

Muse Wonder Full: klog er ikke det samme som god til daglig kode

Morgan Lintons VulcanBench-kørsel giver et køligere vibe check: Wonder Full er præcis, men langsom og dyr; han ville selv vælge noget andet til kodearbejde. Modelkortene er vist, mens komplette kørselsdata og traces stadig er på vej.

Morgan@morganlinton

Okay so I managed to squeeze in getting model cards out today for my Muse Wonder Full benchmark with @VulcanBench - enjoy! This, plus all the data from every run will be on the VulcanBench site soon(ish), and all of the traces will be on Github as well. Everything I do with VulcanBench is free and open source. As a truly independent benchmark (it's just me!) I share it all, every single trace from every run, plus the source code for VulcanBench so you can even re-rerun specific tests, or the entire benchmark if you want to. What stands out looking at this is that Astra is fast, but you pay for it. GPT-6.1 Sol is fast and cheap. Opus wins in accuracy, and Muse Wonder is accurate, but expensive and slow. Right now, with the data I have, I don't think I would use Muse Wonder for coding, and I'm not sure that's the market they're going after, yet. That being said, looking at the accuracy numbers, it's clear the Muse team has built a very intelligent model. If they can figure out how to wrangle the time/cost per task, they could very easily become a model usable for daily coding tasks. Live long and benchmark 🖖

♥ 38 · ↻ 1 · 💬 5 · 🔖 2

https://x.com/morganlinton/status/2111547916896895343

Én database pr. agent — uden nabokrig

Supabase får et lokalt projekt pr. worktree, så én agents database-reset ikke sprænger en andens tests. Den eksperimentelle CLI-motor kan også køre uden Docker; Supabase måler cirka 290 MiB i tomgang mod tidligere 2,6 GiB.

Supabase@supabase

People have been asking to run Supabase locally without Docker for a while. You can do that now. 🐳 Plus, every coding agent gets its own local project, so one can reset its database while another is mid test run. https://supabase.com/blog/local-supabase-project-per-agent-with-or-without-docker

♥ 717 · ↻ 50 · 💬 68 · 🔖 408

https://x.com/supabase/status/2110857277494169647

Hurtigere Claude Code? Noget af magien er gzip

Claude Code komprimerer samtalen før upload, og cloud-sessioner starter, mens man skriver. Theo mærker tydeligt forskellen: mindre ventetid kan komme fra transporten, ikke kun modellen. Kedelig teknik. Dejlig effekt.

ClaudeDevs@ClaudeDevs

We keep improving performance across Claude, and you may have noticed it starts responding sooner! Claude Code now gzips the conversation it sends on every turn, so it reaches the model faster. Cloud sessions also start up as soon as you start typing.

♥ 15,089 · ↻ 645 · 💬 556 · 🔖 2,255

https://x.com/ClaudeDevs/status/2111209208012210409

Pele: Mermaid-syntaks uden hele flyttelæsset

Kepanos nye MIT-bibliotek renderer diagrammer til SVG uden dependencies og tilpasser layoutet til smalle skærme. Det skal ind i Obsidian; “op til 500× hurtigere” er projektets eget tal, men playground og kode kan prøves.

kepano@kepano

This is Pele, my new rendering library for diagrams and charts, compatible with the latest Mermaid syntax. - 100-500× faster than mermaid.js - 6× smaller bundle - No dependencies - Easy to theme - Responsive to device width In pursuit of making Obsidian insanely fast and cohesively designed, I always found the Mermaid library to be too big, too slow, and too hard to theme. Pele will be the new renderer in the next version of Obsidian, but you can start using it now. Pele [ˈpɛlɛ] is named after the goddess of volcanoes and fire. It's open source under the MIT license. Try the playground!

♥ 8,203 · ↻ 602 · 💬 197 · 🔖 5,409

https://x.com/kepano/status/2111472054293913749

Til sidst: et helt hav på 61 KB

FFT-bølger, skum og solglimt i én selvstændig WebGPU-side. Ifølge skaberen byggede Claude den på cirka en halv time med screenshotkritik fra subagents — resultatet er sjovere end endnu en landingpage med lilla gradient.

Jurly@jurlycat

This entire ocean is a 61 KB HTML file. Built with Claude Code + Opus 5.5 in about 30 minutes, according to its creator. It runs in the browser using WebGPU: FFT waves, whitecaps, sun glitter, a physically based sunset sky with clouds, and a free-flying camera. No game engine or rendering library. Claude rendered screenshots in headless Chrome, then had separate subagents critique them. The creator gave feedback along the way, mainly on the clouds. The 61 KB build is minified and self-contained. Download it, open it in Chrome with WebGPU support, and explore with WASD.

♥ 2,761 · ↻ 203 · 💬 137 · 🔖 1,975

https://x.com/jurlycat/status/2111306207239557288

Nyhedsbonus

Meta fjerner “train” fra de omstridte virksomhedsvilkår

Efter kritikken er den udtrykkelige tilladelse til AI-træning på tilknyttede websites fjernet fra de kommende vilkår. Crawling til annonce- og erhvervsprodukter står stadig tilbage — en smallere tekst, ikke et farvel til scraping.