Mick.net - Saas 💻 & Aviation ✈️
@mick__net
Followers
1K
Following
6K
Media
310
Statuses
2K
Tweeting about my SaaS, AI, Aerospace, nerdy topics & remote work. Maker of https://t.co/717Ro1wZCw, building https://t.co/TmFQQFnaAT.
Remote
Joined November 2017
🚢Update for . It now tells you the 🔥visitor peak hours of a place! . This makes it easier to decide WHERE and WHEN you should go when you are in a new area 🎉 . Works around the 🌎🏰🏙️
5
0
18
RT @emmalsalinas: creatine is a scientifically proven iq booster but very little gets to our brains when we consume it orally. a bunch of l….
0
1K
0
@AnthropicAI Missing Claude code already! While Claude Code ships, @Cursor with GPT-5-High-Fast just gets stuck in "Found X errors" loops with the plan that Claude wrote.
1
0
0
:-( I tried to reduce my @AnthropicAI Claude Code Max usage to not hit the 5H period cost/token/message limit, but now I'm still limited because I used CC for more than 5H and have to wait 2.5H. I pay for Max to usage extra tokens, but then there is still a time usage cap.
1
0
0
Pretty neat of @WisprFlow to automatically tag the right files in Cursor when using Speech to Text 🎉
0
1
9
GPT-5 is often perceived to be a small update, but with long horizon tasks (like playing pokemon) it is a big improvement over O3.
GPT-5 Plays Pokémon Crystal - Last Update 🔥. Annnnnnd it’s done! GPT-5 beat RED, the final boss, in 9,517 steps vs. 27,040 for o3 - a huge improvement. It went in under-leveled, but the strategy was strong enough to take RED down easily. Well done, @OpenAI GPT-5 feels like a
0
0
2
Updated instructions: . - Use Codex w GPT-5 'High Reasoning' to review/ debug Claude's work. Used Gemini CLI already to outsource token-rich research for Claude's planning. Any other suggestions?.#FrontierHiveMind
1
0
2
Who's building a CLI tool that uses Claude Code, Gemini CLI, Codex (GPT-5) CLI all at the same, reviewing/collaborating on each other's work?. I'm using Gemini CLI as Claude Code MCP, but I feel more can be accomplished.
0
0
0
Funny how the media/influencers often cite research to make them sound trustworthy and meanwhile they completely fabricate their own conclusion, which sometimes opposes the cited research 🤓.
@DefiantLs The famous blind auditions study supposedly showing sex discrimination is false by the way. It has no significant findings. It was hyped by people and the media who like the conclusion.
0
0
0
Grab also build by some indiehackers? 🤭.
$GRAB users in Singapore and Malaysia were hit with $1,000-plus fares Wednesday after a glitch caused prices to spike nearly 100x the norm. Grab said it was a temporary display issue, with fares back to normal by 12:20 p.m. local time. The company apologized for the error.
0
0
1
@CtrlAltDwayne Correct way to set reasoning effort to high if with the following command (added "-c"):. codex --dangerously-bypass-approvals-and-sandbox -m gpt-5 -c model_reasoning_effort="high".
0
0
0
Time to try out some new Claude Code competition: Codex + GPT-5.
because the codex cli docs suck and there's no easy way to configure stuff, run this: . codex --dangerously-bypass-approvals-and-sandbox -m gpt-5 model_reasoning_effort="high". this is giving me better results than claude rn. like actual god tier results for gpt-5
2
0
4
Smart: Benchmark that can’t be hacked.
🔮 Introducing Prophet Arena — the AI benchmark for general predictive intelligence. That is, can AI truly predict the future by connecting today’s dots?. 👉 What makes it special?. - It can’t be hacked. Most benchmarks saturate over time, but here models face live, unseen
1
1
2
RT @ProphetArena: 🔮 Introducing Prophet Arena — the AI benchmark for general predictive intelligence. That is, can AI truly predict the fu….
0
149
0
RT @jankulveit: Being human in an economy populated by AI agents would suck. Our new study in @PNASNews finds that AI assistants—used for e….
0
37
0
Occasionally you need to tell it to continue (if it prematurely stops).
0
0
0