Osama Romoh

@cjzafir Impressive! What’s the context window?

Our first model Mac-1 6.6B beating 3 giant models. - Haiku 4.5 - GPT 5.4 mini - Gemini 3 flash Running this model on my Macbook M3 24GB. (model takes only 7GB RAM) It searches web, call tools, ask follow-ups, tell jokes, find contacts, search files, write emails, book events, write notes, set reminders and so much Siri can't do. Read again, a 6.6B model. Will share full 2000+ scenario test results & benchmark scores in 2 days.

@javilopen Releasing new iOS with new wallpapers

Seriously, where is Apple?

@ClaudeDevs How does the model actually weight those mid-stream instructions against the cached ones when it comes to long context?

With Opus 4.8, you can add system instructions mid-conversation without breaking the prompt cache. More cache hits means lower cost and latency for your API requests.

@rohit4verse Well.. he still talks to one Claude.

Claude with API access to your Google Ads account is a new media buyer who: - Works 24/7 and never asks for coffee - Will pause a $200/day campaign at 2am because the CPC "looked weird" - Doesn't know your client's quarter ends Friday - Has zero fear of restructuring an ad group you spent two weeks building lol - Will email you a confident weekly report even when half the conversion data is from a tracking outage What else?

@elonmusk Try Claude Code

Try Composer 2.5

Whoever says that GPT 5.5 is better than Opus 4.7 these days.. I gotta say, that's 100% accurate.

Is SuperGrok Heavy worth the $300/month? I’m keen on testing out Grok Build tbh.

Building an AI agent used to feel hard. Here's what I actually do now: – Find a YouTuber who's an expert in the thing. – Drop a few of their videos into Claude. – Ask Claude to write a prompt that mimics how they think. Done. 3 minutes. Free. Better than 90% of "prompt templates" people sell.

@theo What’s your go-to alternative?

I have cancelled my subscription.

I can't help but feel personally burned by the Claude Code changes announced today. We put so much work into wrapping the (atrocious) Claude Agent SDK in T3 Code. It was the ONLY path they supported, so we made it work. It was hell. Now our users are getting their rate limits cut by 40x, despite us doing everything right. I listened to the Claude Code team. I had my issues with their direction, but I trusted them and took them at their word. I will never make that mistake again. Until we see significant change, it is safe to assume any statement from an Anthropic employee is a lie on a timer. The rug will be pulled, no matter how many promises are made beforehand.

@M2Fauzaan I like how simple your app is, but yeah do ship that navigation view.

I can’t decide if I want to ship this. I want to keep this simple but...

I need one for trading 0DTE SPX options xD

OpenAI and Microsoft broke up. The impact of this is massive and I don't think enough people understand why. Put a lot of time into tracking the history of the deal to help you all understand 🫡

Most people using Claude Code touch about three features. I use twenty. Every day. Daily means daily. I ranked them by how often they actually save me hours, and the top of the list isn't close. ✅ Subagents are number one. Parallel work in clean context windows. Nothing else comes near. ✅ Skills are number two. Packaged expertise loaded on demand, so I'm not retyping the same instructions every session. ✅ Then slash commands at three. Repeat workflows collapsed behind one keystroke. ✅ Plan mode at four. I see what's about to happen before any action is taken. ✅ CLAUDE.md at five. The only way I've found to make project rules actually stick. The rest of the list is the muscle behind a real workflow. Honestly, if any of these dropped off the list, I'd notice within an hour. 😁 Full ranking in the graphic. 👇 Feel free to save this post or repost it to your network.

Three frontier models dropped in the last few weeks: GPT-5.5, Opus 4.7, and Gemini 3.1 Pro rolling out wider, and LinkedIn turned into a benchmark trading floor overnight lol. Nobody actually building with these things seems to care which one "won," because there's no best model. There's only what you're trying to do this week. My stack right now: * GPT-5.5 runs the agentic stuff, computer use, multi-step jobs, anything I want chaining tools without me babysitting it. * Opus 4.7 handles the code; it's the first one I trust on production work without rereading every line. * And Gemini 3.1 Pro picks up whatever's cost-sensitive like bulk image work, long video transcripts, anything where the bill gets ugly fast. Different jobs, different tools. The "Claude is dead" / "GPT won" crowd mostly isn't building anything that has to work on Monday. What's doing what in your stack this week?

@M2Fauzaan We can see live changes in preview mode?

Anyway, I fixed it. I built a tiny native app that lets you open any Markdown file on Mac. > Just press Spacebar → instant clean preview > Watch your AI agent write in real time > Open your favorite Markdown editor to edit Install once and literally forget it exists.

Nice work.. any plans for dark mode?

Anthropic just pulled Claude Code from the Pro plan.. is this true? 😱

Grok 4.3 dropped on Friday with basically no press. One low-key post from Musk, no release notes, no blog, no livestream. It just showed up in the model selector for people paying $300 a month. $300 a month, by the way. That's SuperGrok Heavy, the top tier. Regular SuperGrok at $30 can see the model in the dropdown but can't actually click it. What xAI shipped is a half-model. It's 0.5T parameters, not the full 1T version that's still training. Musk confirmed that part himself. So Grok 4.3 right now is half of what Grok 4.3 is supposed to be, priced 10x higher than consumer Grok, with zero marketing telling you any of that. Companies don't quietly ship products they're proud of. A launch goes stealth when the benchmarks came back mid and you'd rather have the sub money than the comparison charts. Compare it to OpenAI this week. GPT-Rosalind for life sciences, partnered with Moderna and the Allen Institute, full press push, the whole accelerate-drug-discovery framing. Whether you buy that or not, it's what a company that wants eyes on its launch does. xAI is shipping like a company that doesn't. Anyway, time to test Grok 4.3 beta. 😄

Goodbye Canva!