taa၊||၊ DeSpeed ၊||၊

airdrop hunter..

[Google Gemma @googlegemma]

Deploying DiffusionGemma-Jev (djev) just got a lot easier. You can now spin up a Jev API-compatible endpoint on Google Cloud Run using a single command.

Performance is solid: ~35-60 ms for single step latency and batch@32 is ~100-123 requests/sec.

It's a straightforward way to experiment without needing your own GPU. Runs at roughly $3/hr and drops to $0 when idle.

Get the code and instructions here: https://t.co/E3agzrW2hZ

[rody @0x_rody]

Jev has been exploding in popularity recently.

If you already have access to the Jev API but aren’t sure how to start experimenting with it, just copy this checklist:

1. agent-desktop
Desktop automation. Read the system's accessibility tree, judge which button, menu, or input field to click next. https://t.co/ZtSEjUSPBF

2. typesafe-mario
Have Jev play Super Mario. No screenshots—just read the structured state in the emulator's RAM, then decide to run, jump, or dodge. https://t.co/GHttIjWQ3p

3. jev-drone
Use Jev to control a drone. The underlying flight control still handles stability and safety; Jev just does higher-level judgments like climbing, braking, and navigating obstacles. https://t.co/z0lYh9ykJq

4. OneVOneJev
1v1 FPS in the browser. Every decision tick, judge movement, view angle, aiming, firing, and jumping. https://t.co/aJiU0aaNcI

5. jev-trader
High-frequency market making on Monad testnet. Jev judges the next buy or sell based on spreads and trade direction, with model latency around 81ms. https://t.co/DaDRIrkJpO

6. Prism
Doesn't directly have Jev place orders. It judges states like toxic flow, market pressure, mean reversion, etc., then hands off to the original strategy. https://t.co/aim9lGRAP8

7. neo4jev
Stuff Jev into a knowledge graph. At each node, judge the most worthwhile edge to take next, then follow it all the way. https://t.co/9h0KXKdWj9

8. jev-curate
Use Jev to screen training data. For JSONL / Parquet, first judge quality, relevance, and risk, then decide which ones go into the next training round. https://t.co/yYV6aEdUtG

9. Canny
Prevents Coding Agents from stubbornly claiming they're done. Look at tool outputs, code diffs, and test results, then judge if the completion claim is reliable. https://t.co/H4jFT8hV0E

10. killmyidea
Input a startup idea, and Jev scores it from multiple angles, finally giving you KILL, FIX, or SHIP. https://t.co/XWo7JPOb6y

Copy these complete Jev blueprints - then read full Jev setup below ↓ ↓

[Dewaldt Huysamen @GodsBoy7777]

Top 10 Hermes skills ranked by Github stars thanks to reddit thread.

https://t.co/bHgRjFguT4 - An agentic skills framework & software development methodology that works.

https://t.co/FXpgttwhrv - Makes your AI agent think like the laziest senior dev in the room. The best code is the code you never wrote

https://t.co/6LRxTj6nIB - Turn any codebase, with its docs, SQL schemas, configs, and PDFs, into a queryable knowledge graph

https://t.co/mfPGpI65JU - Viral skill + proxy for coding agents that cuts 65% of tokens by talking like a caveman.

https://t.co/EK7YebfnwU - Turn any code into an interactive knowledge graph you can explore, search, and ask questions about

https://t.co/Rbby599mcc - AI agent skill that researches any topic across Reddit, X, YouTube, HN, Polymarket, and the web - then synthesizes a grounded summary

https://t.co/k2yWmbc2ns - A skill to stop your coding agent from burying the answer. ADHD-friendly output.

https://t.co/cA5E9uTlsd - AAS Core is the local, agent-first control plane for complete catalog discovery, agent-owned selection, stack validation, and planning, backed by 2,400+ agentic skills.

https://t.co/Qiv7Um1CFM - 165 ready-to-use validated skills plus 100+ scientific databases covering biology, chemistry, medicine, and drug discovery

https://t.co/RL1p8VvHm7 - Editorial diagram design for Claude Code, Codex, and Pi. Self-contained HTML + SVG. No shadows. No Mermaid slop.

[SuSu_酥酥👅 @NFT_Chen]

🤨还有谁不服？Laya 出马 Jev 直接变成弟弟！

16GB MacBook Air，本地 Laya 45ms 决策；云端 Jev 约 300ms。

Laya 比 Jev 速度快 7–11 倍，而 Laya 成本 $0 vs Jev 按 token 收费，关键是 Laya 权重还开源！

俄罗斯方块里这不叫评测，这叫手速碾压。

同纬度的 System One，快的那个先清行。

本地运行 AI 模型 -> https://t.co/uE21YROWs8

#Laya #Jev #LocalAI #开源AI

https://t.co/1CYXsnsbm3

[Bilal Khan @Bk23544]

JEV is crazy!!!

I got access, so naturally I had to try it with the Three.js Grassworks demo.

I built an AI Scene Director that takes a simple prompt and changes a whole set of scene and grass parameters to customize the environment based on your instructions.

And it’s FAST. Really damn fast.

You can literally describe what you want and watch the entire environment transform.

Should I add this in the live Three.js Grassworks demo?

#jev #typesafeai #threejs

[Dhairya Karekar @dkare1009]

A Russian guy has discovered a method to learn anything ten times faster using AI! It involves NotebookLM, Gemini and Obsidian.

[MIKE @mikenevermiss]

this is f**king insane.

an open source team just released a FREE tool that can turn one video into content for almost every platform.

it already has 11.4K stars on GitHub and keeps climbing.

it lets you:

→ download or upload any video
→ transcribe, translate, and dub it in 100+ languages
→ automatically turn landscape videos into vertical ones
→ generate the cover art too

i threw a 46-minute video at it and it subtitled the whole thing in one run with zero manual fixes.

endless video content for YouTube, TikTok, Bilibili, Douyin.

save this. you’ll need it.

[Visharad @KashyapVisharad]

i made Jev (@typesafeai) play clash royale.

Qwen 3.8 27B on @cerebras reads the battlefield, OpenCV reads cards + elixir, and Jev picks the card + placement.

fresh account, starter deck. it won multiple games against REAL people. i'd be so annoyed

[Nick Khami @skeptrune]

you can make any open source model behave like jev with just a bit of inference engineering.

it's shockingly easy. to prove it, we built a new endpoint we're calling deepseek-v4.1-flash-jev. see the demo below. here's how it's done:

sglang (an inference engine) offers a scoring endpoint in addition to the normal generation one. in scoring mode, given an input & set of possible answers, it forces the model to produce probabilities for each one. example:

> input: what is most common letter in abcccde?
> possible answers: a, b, c
> output: (c, 0.9), (b, 0.0.5), (a, 0.05)

getting the above behavior instead of streamed output is as simple as using sglang's /v1/score endpoint instead of /generate. there's just one other trick required.

for deepseek, you have to add a closing think tag before the response. this forces a direct answer instead of a reasoning trace. if you want reasoning, you can do that too, but imo that makes things too slow to be worth it.

dsv4.1 flash is not as good as jev, but if we had enough spare compute to experiment with this same approach for a larger model then i think the decision quality would be at least as good, if not better.

also, somewhat unrelated, i think decision-making models kill all prospecting & sourcing work. i would have absolutely killed to have jev or similar when i was recruiting @mintlify. absolutely incredible.

[shiv @sxhivs]

jev has changed computer-use forever.

i built a computer use tool with jev that costs basically nothing to use, and has sub-second latency.

it can even decide what to type, and doesn't need an LLM.

open source. link below. no LLM required.

[Kai @hqmank]

Combined Jev with Playwright-controlled Chrome to build jev-browser, a general browser automation skill.

Works across agents. Two demos in Antigravity CLI and Codex: finding related articles, then job openings on a site.

Both ran fast. This could handle many more browser tasks.

[Morty @0xMortyx]

Google Senior Engineer Annie Wang just dropped a free hands-on course on Graph RAG agents, from scratch:

Setup → Graph Schema → Embeddings → Hybrid Search (RRF) → ADK Agent → Deploy

0% → 0:00 – a Graph RAG agent with ADK
25% → 13:01 – Build the graph schema and visualize your data
77% → 39:16 – Hybrid search + RRF
84% → 43:05 – Wiring up the ADK agent: runner, session service, memory service
100% → 51:06 – Final demo and deploying the agent with FastAPI

This is one of the few guides that walks through real code and architecture, from an empty project to a working agent.

Watch it, then build it yourself

[codila @0xCodila]

Google just released free 2-hour course on full Graph & Loop engineering: 1 prompt → 100 agents → loops → graphs from 0% to 100%:

0% → 0:35 - Graph engineering from scratch
30% → 31:17 - build your first agents graph
45% → 43:40 - run hundreds agents in parallel
75% → 1:04:58 - Loop engineering: route, check, repeat
100% → 1:30:09 - self-improving graphs that work while you sleep

most people keep stuffing instructions into one agent - the real upgrade is a graph that routes, checks, and rebuilds the workflow at runtime

learn graph from Google, ship your first - then unlock the complete system design below ↓

[Google Antigravity @antigravity]

Did you know you could do a split agent view?

Check out these quick Antigravity tricks⬇

[Lonely @Lonely__MH]

🎉视频 AI 剪辑省 Token 组合： Antigravity + ChatCut ! ！

gemini 3.8 flash 的速度 和 token 消耗

得天独厚，为所欲为！！

比 Codex 操作快多了！太爽了！

🔥方法如下：
将提示词发送到 Antigravity 对话框即可
---
连接到此电脑上的本地 ChatCut Desktop MCP 服务器。

使用以下确切命令或可执行文件注册名为 `chatcut_desktop` 的 stdio MCP 服务器：
"/Users/{你的用户名}/Library/Application Support/ChatCut/chatcut-mcp"

注册后，请验证 `chatcut_desktop` 已启用且可被发现。然后启动一个新的 Agent 对话，以便使用该连接。
---

灵感来源： @johnAGI168

[火山哥🕊️ @huoshan007]

兄弟们爬虫界要翻天了，这个太牛逼了。

browser-use 我挖出来的时候还没这么火，刚看了眼，已经 9.5 万星了。

两位苏黎世 ETH 的研究员搞的，MIT 开源。

说白了就是让 AI 像真人一样操控浏览器：

点击、滚动、登录、填表、换页面、从没见过的网站里抓数据。

我自己试过以后才敢说，这东西对普通爬虫挺狠。

很多传统爬虫摸不到的页面，尤其是动态渲染、要交互、要登录的地方，它直接让 AI 上手点。

最关键的不是“能爬”，而是它开始会判断页面该怎么走。

这点有点吓人。

链接：

https://t.co/S7ELiYf4SO

[witcheer @witcheer]

a Hermes Agent community member built a bot whose only job is making other specialist bots.

indeed, a new bot should be more than a copied profile with a personality file pasted in.

botmaker interviews you about the job, drafts the new bot's SOUL for your sign-off, scaffolds the profile with no inherited skills, then tests the new bot in its own chat before writing it into your fleet docs. if the job you describe is fuzzy, it stops and asks instead of building.

https://t.co/SeDpszHFON

[Harman @itsharmanjot]

You can now scrape the web with AI agents in 85ms.

Someone built in Rust what Chrome engineers couldn't optimize in years.

Obscura is an open-source headless browser engine for web scraping and AI automation.

A 70MB binary that boots instantly, loads pages in 85ms, and runs on 30MB of RAM.

The Google equivalent needs 300MB of disk, 200MB of memory, and 2 seconds to start.

The real trick is stealth mode.

Per-session fingerprint randomization across:

> GPU, canvas, audio, battery
> 3,520 tracker domains blocked
> navigator.webdriver fully masked
> Native functions patched cleanly

No detector can catch it.

The engine runs real JavaScript through V8 and speaks the Chrome DevTools Protocol natively.

That means it works as a drop-in for Puppeteer and Playwright.

You swap the binary, your code stays untouched.

One developer shipped this alone. Free to use.

[Tom Dörr @tom_doerr]

Automate real-browser interactions and extract web data using BrowserAct skills.

https://t.co/TMghhL1iKv

[MIKE @mikenevermiss]

DOWNLOADING ANYTHING FROM THE INTERNET JUST GOT ABSURDLY EASY.

someone built an open-source tool called Cobalt that lets you download videos, audio, photos, and GIFs from practically any website.

all you have to do is paste the link and it handles the rest, no ads, trackers, paywalls, or unnecessary steps.

It already has 42.3K+ stars on GitHub, which tells you how useful this thing is.

I’ll leave the link in the comments.