Rez

orchestrating reliable agents....

[Rez @creativerezz]

When you start to doubt your idea - remember that DropBox started as a python script written on a bus

[Rez @creativerezz]

@GoogleDeepMind @grok how do I use genie 4

[Rez @creativerezz]

https://t.co/qBKo1hqO9r

[Rez @creativerezz]

@claudeai @grok find suggested prompts to use with Opus 5.5 from Anthropic official docs

[Rez @creativerezz]

Deployed ZCode locally on macOS with GLM-5.3.

Currently outperforms every other open-source model on Terminal Bench 3.0 and Agents’ Last Exam (CLI).

@AutoClawAIer

[Rez @creativerezz]

When you start to doubt your idea - remember that DropBox started as a python script written on a bus

[AutoClaw @AutoClawAIer]

GLM-5.3, the No. 1 open-source model for coding, is officially here.

GLM Coding Plan members can now complete the setup in AutoClaw and get early access to GLM-5.3.

Limited-time benefits:
-Get 1.5× usage quota when using GLM-5.3 and other GLM models through GLM Coding Plan.

GLM-5.3 ranks No. 1 among open-source models on public benchmarks including Terminal Bench 3.0 and Agents’ Last Exam (CLI), delivering stronger and more reliable performance in complex code understanding, long-horizon execution, and multi-step agentic work.

Try GLM-5.3 in AutoClaw—and put it to work on a truly complex task.

[Rez @creativerezz]

@nickywonka told @theSamParr and @ShaanVP every grocery category is getting its own fridge. Pet food already did it. Chocolate did it.

Walmart baby aisle today. They weren't kidding.

[Rez @creativerezz]

a 3x faster (up to 7x on tagged templates), byte-identical drop-in for your cn() utility.

Same API, same output, slightly bigger bundle. Pull it in if you have re-render-heavy UIs.

What it actually is
Re-implements clsx + tailwind-merge with heavy optimizations while guaranteeing identical results. The standout feature is the tagged template for
•Standard cn: ~2.5M ops/s
•cnfast: ~6.6M ops/s
•Geometric mean across 65 real workloads: 3.17x
•Biggest wins on data grids, virtualized lists, live dashboards, anything calling cn thousands of times per second.
Bundle: 9.04 KB gz vs 8.45 KB. Negligible.
How to adopt (copy-paste ready)
For a standard shadcn lib/utils.ts:
// before
import { clsx, type ClassValue } from "clsx";
import { twMerge } from "tailwind-merge";
export const cn = (...inputs: ClassValue[]) => twMerge(clsx(inputs));

// after
export { cn } from "cnfast";
Or one-liner:
npx cnfast migrate
shadcn users:
npx shadcn@latest add aidenybai/cnfast/cn
Caveats
•If you’re doing custom extendTailwindMerge for prefixes/MFE stuff, test it. The exports include twMerge/twJoin, but custom wrappers need verification.
•Static/low-re-render pages won’t notice much. Hot paths will.
•Still early (shipped yesterday). Watch the repo for a couple days if you’re paranoid about prod.
Take
This is the exact kind of micro-optimization Aiden does best. cn is one of those “invisible tax” functions that adds up in any shadcn/Next.js codebase with conditional classes. Low risk, easy rollback, real upside where it matters.

Do the swap in one of your heavier UI projects (Mermaid Studio tables or whatever has lists), run a quick before/after on a render-heavy component, and measure.

Takes 2 minutes and you will understand wh

[Z.ai @Zai_org]

Introducing GLM-5.2: Frontier Intelligence, Open Weights

- Significant improvements in coding and agentic tasks
- Strong long-horizon capabilities with a 1M context window
- Two levels of reasoning effort: GLM-5.2 (max) pushes the limits, while GLM-5.2 (high) strikes a strong balance between performance and token efficiency
- MIT-licensed open weights
- Same API pricing as GLM-5.1

Tech Blog: https://t.co/LAsxUdN0JZ
Weights: https://t.co/g0A1C4UWx4
API: https://t.co/Kc3E22cbN7
Coding Plan: https://t.co/Nk8Y98HNhU
Chat: https://t.co/WCqWT0qCQb

[Kimi.ai @Kimi_Moonshot]

Meet Kimi K2.6: Advancing Open-Source Coding

🔹Open-source SOTA on HLE w/ tools (54.0), SWE-Bench Pro (58.6), SWE-bench Multilingual (76.7), BrowseComp (83.2), Toolathlon (50.0), Charxiv w/ python(86.7), Math Vision w/ python (93.2)

What's new:
🔹Long-horizon coding - 4,000+ tool calls, over 12 hours of continuous execution, with generalization across languages (Rust, Go, Python) and tasks (frontend, devops, perf optimization).
🔹Motion-rich frontend - Videos in hero sections, WebGL shaders, GSAP + Framer Motion, Three.js 3D.
🔹Agent Swarms, elevated - 300 parallel sub-agents × 4,000 steps per run (up from K2.5's 100 / 1,500). One prompt, 100+ files.
🔹Proactive Agents - K2.6 model powers OpenClaw, Hermes Agent, etc for 24/7 autonomous ops.
🔹Claw Groups (research preview) - bring your own agents, command your friends', bots & humans in the loop.
-
K2.6 is now live on https://t.co/YutVbwktG0 in chat mode and agent mode.
For production-grade coding, pair K2.6 with Kimi Code: https://t.co/uvoSJKyGCY
-
🔗 API: https://t.co/EOZkbOwCN4
🔗 Tech blog: https://t.co/9wWvgIQSS3
🔗 Weights & code: https://t.co/Be0hjs2RTP

[Rez @creativerezz]

@HeyGen CLI

[HeyGen @HeyGen]

Your AI agent can now generate and ship videos.

HeyGen CLI is now live.

Run one command and your agent handles it all:
script → avatar creation → video → delivery

All from the terminal. Just your agent and the CLI.

RT + Comment “CLI” and we’ll DM API credits (must follow)

[Rez @creativerezz]

@adelholzerkaese @NandinoAI @ManusAI @grok claude co-openclaw

[Andrew McCarthy @AJamesMcCarthy]

The planet Saturn, captured during it's closest approach to Earth from the top of a parking garage in the heart of Los Angeles.

You can do plenty of astrophotography from a big city!

[Scale @scale_AI]

Introducing SWE-Atlas.

We built SWE-Atlas as the next evolution of SWE-Bench Pro, expanding agent evaluation beyond change accuracy to better reflect the real, interactive workflows that define software development.

Results for Codebase QnA, the first eval under SWE-Atlas that measures how agents understand complex codebases through runtime analysis and multi-file reasoning, are now available.

Top models score only ~30%.

https://t.co/VtoXDkYYL6

[Rez @creativerezz]

@AlfieJCarter claude

[Johann Sathianathen @johann_sath]

default openclaw:

workspace/
├── https://t.co/dD8xK2e6G7
├── https://t.co/HvO3hxrYAE
├── https://t.co/bstTTdPH7K
├── https://t.co/w5c6tyPnPQ
└── skills/

that's it. a chatbot with personality.

my openclaw after 3 weeks:

workspace/
├── https://t.co/dD8xK2e6G7 (customized)
├── https://t.co/HvO3hxrYAE
├── https://t.co/bstTTdPH7K
├── https://t.co/w5c6tyPnPQ
├── https://t.co/6AN4DpUqbH — live working memory
├── https://t.co/N8fXVofhiB — long-term memory
├── https://t.co/joOFuvsTMv — autonomous thinking loop
├── https://t.co/QHJQzDCqXZ — client profiles
├── https://t.co/YtlFVY532X — decision frameworks
├── https://t.co/DGe5V4NpVG — writing voice guide
├── https://t.co/8yLvJBW8ek — startup rules
├── memory/ — daily logs
├── skills/
│ ├── tweet-writer/
│ ├── website-builder/
│ ├── website-dev/
│ ├── script-polish/
│ └── security-auditor/
├── content/
├── consulting/
├── drafts/
└── crm/

this is the difference between a chatbot & an AI employee.

none of this is built in. i added every file.

your setup is your moat.

[Rez @creativerezz]

Cloudflare has helped me save 148 GB in the last month. #cloudflare

[Rez @creativerezz]

Why does @cursor have 30+ @OpenAI GPT model variants? 😅

GPT-5.2 Codex Fast, High, Low, Extra High Fast...
GPT-5.1 Codex Max High Fast, Low Fast, Extra High Fast...

Is this what eval paralysis looks like? 🤔