Make this a habit
We tested Claude Opus 5.5 for a week. Some tips:
1. Set a budget and a stopping point. It won't stop on its own, and it'll eat your weekly limit. Tell it what "done" looks like.
2. Ask for the deliverable first. Given 10 minutes for a run of show, it built a data generator and handouts, then ran out of time before making the schedule.
3. Let it critique itself, then check. It caught things our designer missed, and also rejected 90 percent of its own work for reasons he couldn't trace.
4. Give it notes. Opus 5.5 is a personality fix: It builds on your feedback now instead of arguing with it.
5. Move the point up yourself. Its prose is the most readable we've measured, but it still buries the lede.
6. Point it at visual work. Game UI, 3D, animations, and prototypes were its best one-shot results.
Full Vibe Check: every.to/vibe-check/vibe-che…
DimaMan retweeted
DeepSeek-V4.1-Flash: the BEST Boeing benchmark result I’ve seen from an open-source model (by far). ✈️
It ran for hours with /goal. Where other models often stall quickly, it kept improving: inspect, spot a defect, zoom in, diagnose, fix, repeat. The video really shows how strong it is at sustaining that process over a long horizon (and how good it was at identifying what to work on). Impressive.
Harness: Claude Code
Prompt: "/goal create the most realistic boeing 747 using THREEJS - use your vision capatilities to create a self verifiable system, enter a loop until you are 100% satified about the result (you can build a camera system to inspect each angle)"
It'is so obvious that this is the future of IDE / ADE
Introducing Projects, a new way of working in Cursor.
Rather than creating a chat for every task, you work with a coordinator agent in a single, persistent thread.
Like @bot, your agent is always on, proactively manages work with subagents, and improves over time.
and giving it for free on the 20$ sub, and not fear mongering 200$ subs, like others
Astra 3D + Math capability remind me of old documentary about the boy who had incredible ability to see the numbers in shapes.
OAI must have invited some new LLM architecture that is inspired by this.
youtube.com/watch?v=OyDg4p9Y…
DimaMan retweeted
one very cool development with gpt 6 astra is that you can change reasoning effort mid-convo without invalidating the cache. start with astra light (roughly equivalent to sol medium), then jack the thinking effort up if you see the task can benefit from more, all without worrying about having to rewrite the complete context cache.
it is so obvious,
OxAlpha is Curosr Composer 3
based on GLM 5.3, RLHF with Fireworks, served on xAI in USA
Composer 2.5 was big success based on K2.5
what a generational fuckup by @emotiv . they are building high-tech products for decade, but cant get their customer support in order...
at least we can extract a lesson here as founders.
Dear @emotiv I'm really losing my patience here, I ordered Emotiv EPOC X and paid 1169.05 $ on May 15, 2026. Opened a ticket in June, another one in July.
Your automated response has always been "Thank you for contacting EMOTIV. Your request has been received and is being reviewed by our service team. We will do our best to get back to you within 12 hours. Thank you for your patience and understanding!"
I've never get back anything after this.
I'd like a refund now, how can I proceed? Does anyone have contact with anyone working in this company? I was planning magic with their ECG reader, but I'll search other paths.
DimaMan retweeted
The biggest update to PlayStation VR2 Toolkit yet, and more, is here, v1.0.0 (experimental #1)! Today, we have released the PS VR2 jailbreak (unlocking eye tracking camera feed, headset vibration) so that you can use Babble or headset vibration for headpats.
More info below 👇
DimaMan retweeted
Claude Opus 5 one-shotted this game.
EVERYTHING you see in this demo is custom code... not a single external asset was used.
AI games are going to be amazing.
(sound on)
