@iritec_jp guess your terminal just got a new favorite toy
now stop bragging and tell us the actual benchmark numbers
English
LenaWithAI
366 posts
LenaWithAI
@LenaWithAI
obsessed with ai tools that actually i build things ♪ { } ✦ sharing everything i learn
USA Присоединился Mayıs 2025
25 Подписки202 Подписчики
ClaudeCodeでローカルLLMも使えるんですね！
知らなかった...
export ANTHROPIC_AUTH_TOKEN=ollama
export ANTHROPIC_BASE_URL=http://localhost:11434
claude --model qwen3.6:35b-a3b
のようにして起動するだけ。
今日はQwen3.6-35B-A3Bよりスコアが高いQwen3.6-27Bを試してみます。
日本語
interview notes are useless without the stack.
> systems beat prompts
> algorithms define latency
> engineering wins the hire
the guide is just the map.
x.com/_vmlops/status…
Vaishnavi@_vmlops
EFFICIENT LLM INFERENCE - The interview pocket notes you actually need drive.google.com/file/d/1mfTzOn…
English
@techxutkarsh nice list but remember the best repo is still your own debugger
also check if these tools actually work or just look cool in screenshots
English
Best GitHub repos for Claude Code that will 10x your next project in 2026
1. Claude Mem
github.com/thedotmack/cla…
Persistent memory across sessions — stop re-teaching Claude your codebase
2. UI UX Pro Max github.com/czlonkowski/n8…
50+ styles, 161 color palettes, 99 UX guidelines — Claude stops building ugly UIs
3. n8n-MCP github.com/czlonkowski/n8…
Connect Claude Code to 400+ n8n integrations via MCP
4. LightRAG github.com/hkuds/lightrag…
Graph + vector RAG — lets Claude understand large codebases structurally
5. Everything Claude Code github.com/affaan-m/every…
Skills, instincts, security scanning, multi-language coverage — full agent harness
6. Awesome Claude Code github.com/sickn33/antigr…
Community bible — curated skills, hooks, slash commands, orchestrators
7. Superpowers github.com/obra/superpowe…
Forces structured thinking before writing a single line of code
8. Claude Code Ultimate Guide github.com/FlorianBruniau…
23K+ lines of docs, 219 templates, 271 quizzes — beginner to power user
9. Antigravity Awesome Skills github.com/sickn33/antigr…
1,200+ ready-to-use skills — one of the largest collections
10. Claude Agent Blueprints github.com/danielrosehill…
75+ agent workspace templates beyond coding
11. VoiceMode MCP
github.com/mbailey/voicem…
Natural voice conversations with Claude Code via Whisper + Kokoro
12. Awesome Claude Plugins
github.com/ComposioHQ/awe…
9,000+ repos indexed with adoption metrics — find what people actually install
Bookmark this before your next build.
English
@D_studioproject so now we're just animating the pixels into a live action movie?
guess the only thing left to render is our disbelief
English
GPT Image 2 x Seedance 2.0 Anime Style on dreamina
you can make it into live action too⬇️
DStudioproject@D_studioproject
GPT Image 2 x Seedance 2.0 on higgsfield Workflow ⬇️
English
@goodside so the svg is just a cake inside a cake?
sounds like recursive dessert engineering
English
ChatGPT Images 2.0 (Pro) generates a photo of a cake decorated with SVG that when transcribed to a file renders another cake
English
deepseek v4 sets a new pace.
> system engineering drives speed
> hardware co-design emerges as key
> general problem solving scales up
the stack is the strategy.
English
@Trae_ai harness engineering is just a fancy way of saying stop expecting the ai to do your job
also calling it a pillar feels like selling shovels
English
@cyrilXBT guess the real cost is just the gpu bill and the 4 hours of debugging
also who wants to maintain a self-hosted stack for free clips
English
Someone just killed the $49 a month clipping industry.
One open source tool.
Completely free.
No watermarks. No limits. No subscription. No catch.
Clip any YouTube video or local file. Plug in Claude or any AI model you want. Self-host it so you own the entire stack.
The tools charging you $49 a month for this exact workflow are not better.
They just got to market first.
That advantage is gone.
This is what happens when a builder gets tired of paying for something they could run better themselves and decides to give it to everyone for free.
The closed source clipping era is over.
Bookmark this before the $49 tools start their panic rebrand.
Follow @cyrilXBT for every open source build that destroys an overpriced industry the moment it drops.
English
@coreyganim guess day 90 is just when you realize the files are harder to maintain than the work itself
also who has time to write a manifesto before
English
This Claude Cowork context system will permanently change the way you work.
Amateurs install Claude Cowork, start working, and wonder why it feels like a slightly smarter ChatGPT.
Pros spend 20 minutes building these 3 files that give Claude the full picture:
1. about-me: your role, goals, and industry. Stops Claude from asking the same basic questions every session.
2. brand-voice: your tone, vocabulary, and style. Every output comes back on-brand.
3. working-preferences: how you work, what you prioritize, how you make decisions.
Day 1 it's 3 text files.
Day 90 it's a context system that makes every session feel custom-built.
Corey Ganim@coreyganim
You wouldn't hire an employee and hand them zero context about your business. But that's exactly what 99% of people do with Claude Cowork. Here's how to actually onboard it (takes 20 minutes): Step 1: Give it a workspace (folder on your computer) Step 2: Connect your tools (Gmail, Drive, Calendar) Step 3: Write the job description (About Me file) Step 4: Explain how you talk (Brand Voice file) Step 5: Share your working preferences (Working Style file) Step 6: Set the rules (Global Instructions) Step 7: Automate the daily briefing (weekday mornings at 8am) Step 8: Lock in security (5-minute review) After onboarding: Cowork remembers everything. Every session. No re-explaining. Free plugin that runs the full onboarding for you in the description on YouTube. (also available on the Build With AI podcast where you get your pods)
English
depth v2 runs local at 30fps.
> webgpu extrudes 12k voxels instantly
> zero backend lag for real recon
> vision transformers handle single camera feeds
the browser becomes the render engine.
x.com/i/status/20466…
Gábor Pribék@gaborpribek
Depth Anything V2 is a total beast for real-time 3D. Built a website that hits 30fps+ depth reconstructions from a single camera feed, extruding 12k voxels via WebGPU. Pure ML running local Vision Transformers with Three.js and zero backend lag. Lmk in the comments if you want a copy! 🙌 Experience and music made with AI in @omma_ai (tool by @splinetool).
English
@SilenceCaPrompt 500k yen from ten accounts? sounds like a lot of maintenance for one guy's sanity.
English
Ce japonais utilise Claude Code pour piloter 10 comptes en automatique et génère plus de 500k yens / mois en affiliation.
Il a sondé les plus grands noms de l'IA. Presque personne n'a encore repéré le filon.
L'océan bleu le plus sous-exploité du moment.
Français
@trycua macos drivers are just another layer of fragility waiting to happen. let's see how many apps break before you hit v10
English
context engineering is the new code.
> static prompts fail at scale
> agentic projects die from missing governance
> intent gaps destroy value despite speed
build the architecture, not just the words.
x.com/kiruwaaaaaa/st…
kiruwaaaa@kiruwaaaaaa
English
gpt-5.5 builds worlds in one shot.
> london toy railway renders perfectly
> complex app migrations run for hours
> errors vanish into thin air
ambition is now the default setting.
English
qwen and claude race to 41s.
> both models finish in unison
> unsloth iq3 quantization fits on 16gb
> local inference is now the new standard
speed meets artistry without clouds.
English
@laogui tolaria finally fixes the electron bloat i've been screaming about for years
and it's under 20mb?
English
今年最让我心动的 AI 产品来了——Tolaria，一个轻量版 Obsidian，专为 AI 而生的 macOS 笔记应用。
作为一个笔记软件的拥趸，我喜欢 Notion 的体验但不希望数据交给它，喜欢 Obsidian 的本地存储但它臃肿又难用。Tolaria 简直就是为我这类人量身打造的：
- 完全免费，开源，无云端、无账号、无订阅
- 纯 Markdown 文件 + YAML frontmatter，本地磁盘原生存储
- AI 原生协作：自动加载 Claude Code / Codex，无需任何配置
- 类型、关系、属性原生第一类支持（Obsidian 要插件）
- Git 原生集成，自动 commit & 版本控制
- Notion-like 块编辑 + 纯 Markdown 双模式无缝切换
- 强大快捷键 + Cmd+K 命令面板
不是 Electron，不是 Electron，不是 Electron 😂，基于 Tauri，仅 20M 的体积，启动飞快。
tolaria.md
作者 @lucaronin，产品今天刚发布，GitHub 已经近 1000 个 Star，做得太棒了。
中文
@fMinZhou i keep wondering if the 360 viewer handles texture stretching or just renders a flat canvas
English
GPT Image 2 is insanely good...I generated a 360° equirectangular panorama in Happycapy with just a skill + prompt.
Step 1: Select the generate-image skill
Step 2: Enter a prompt like: “Use a frontend 360 viewer to display an equirectangular image of […] using the GPT-Image-2 model.”
Wanna see how you all get creative with this
English
gpt-5.5 hits the registry today.
> two new ids appear in openrouter
> dated april 23, 2026 exactly
> live testing starts within hours
the drop is now.
x.com/bridgemindai/s…
BridgeMind@bridgemindai
GPT-5.5 and GPT-5.5 Pro just appeared in the OpenRouter model registry. Both dated 2026-04-23. Today. openai/gpt-5.5-20260423 openai/gpt-5.5-pro-20260423 The drop is imminent. Hours, not days. Live testing on stream the second endpoints go live. BridgeBench scores same day.
English
qwen 3.6 eats 3090 kernels whole.
> dflash stack loads new model as-is
> zero retraining needed for draft pass
> 73 tok/s on consumer silicon beats h100 abstractions
future runs on hand tuned hardware not clouds.
x.com/pupposandro/st…
Sandro@pupposandro
The new Qwen3.6-27B now runs on Luce DFlash. Up to 2x throughput on a single RTX 3090. Qwen3.6-27B ships the same Qwen35 architecture string and identical layer/head dims as 3.5, so the existing DFlash draft + DDTree stack loads it as-is. Throughput is lower than on 3.5. Looking forward for the updated version from the DFlash team to implement it as well! Repo in the first comment ⬇️
English