← Back to leaderboard
ziwenxu_

Ziwen

Verifiedactive
@ziwenxu_

26 | Trying new AI tools, building things, and sharing what’s worth your time. https://t.co/uuEn9gTnE0 https://t.co/X7SBufnpNs

RosterAI Content Creation & Video GenAI Model News & Releases
18,922 followers · 471 following · Last polled 3h ago
Perf. Score
36.7
Rank #319
Value Score
32.1
#189 of 273 priced
Median views
1.6K
p25 1.0K · trailing 7d
Followers
19K
+1.3% 7d
ER (impr.)
1.25%
eng ÷ impressions
ER (foll.)
0.76%
eng ÷ followers
Est. CPM
$68.54
QT-rate basis
WoW views
-9.8%
▲ rising

Rates & value · est. from 1.6K median organic views

overpriced
Quote tweetbasis
$110
$68.54 / 1K views
Post
—
no CPM
Thread
—
no CPM
Retweet
—
no CPM
Value Score 32.1 · rank #189 of 273 priced$4313.73 / 1K engagementsrate age unknown (pre-audit import)

Trends

Follower growth
Median views / day
Engagement rate
View distribution · trailing-7d posts

Each blue dot is one organic post. The Performance Score ranks by the median (1.6K), not the mean (12K) — so a single viral post doesn't inflate a creator's reach. The p25 floor (1.0K) is the level 75% of posts clear.

Recent posts · ranked by views

PostPostedViewsLikesRepostsRepliesQuotesBookmarksER
The first ever CLAUDE MADE ANIME is out!! It's called Gus & Spark and 25 minute episode. Opus 5.5 MAX worked on it for over 15 hours, and wrote it code by code with 0 AI generated video. - 34K lines of code - 986M tokens, about $370 at API prices I believe Fable 5.5 should be able to produce even higher quality anime! In the future all 3D animated videos can be created by Claude without any AI generated video.frozen6d ago350K2.1K135142591.7K1.18%
Infinite money glitch here. Ghost's new Core AI computer is $3,499. The RTX PRO 4000 Blackwell + AMD Ryzen 5 sells for about $3,400 by itself. So the rest is basically free?? 64 GB DDR5 RAM - about $1K 1 TB NVMe SSD - $250 Total is like $4.5K if you buy the parts separately. https://t.co/11IWzyB9wwfrozen17d ago245K1.2K27
57
8
743
0.84%
For all Claude Max users, Remember to claim your monthly API credits. https://t.co/pdB6Nvmuyp https://t.co/XbgbMB0NInfrozen14d ago185K1.2K5347111.4K1.42%
As a founder, which personal assistant are you using right now? 1. Dot 2. Muse 3. Grok bot 4. Hermes 5. Openclaw?frozen21d ago165K480848191700.69%
Introducing OpenAI "o", your personal assistant!! It showed up on the ChatGPT Pro upgrade screen as "o, your always-on assistant". Always-on means it keeps working after you close the chat. ppl say it runs on a new model called aeon, built for long tasks. Same lane as Grok Bot. I'm already on Pro, so if the leak is right it lands straight in my sub. OpenAI's DevDay is Tuesday. Can't wait to test it out!!frozen26d ago165K4925432152150.49%
@Pexoai_offical has a feature called Mark to fix. It's code review comments for video. You draw on the bad frame and type what's wrong. It redoes that part and leaves the rest alone. Someone ran it on a paper collage and caught stamps that were never in the approved artwork. Six marks. One note said "This stamp wasn't in the original artwork. Remove it." Finished cut below.frozen29d ago112K59403328320.17%
4 hours in and the entire Claude Projects credit is gone. That's the $100 that doesn't count toward our limits. Plus 19% of our weekly usage over 4 /loops.. https://t.co/rJRiZvF3Uvfrozen12d ago102K1907230.03%
what just happened guys. We maxed out claude's max 20x plan in 19 hours. did they cut the usage limits? it should've lasted us about 2 days. https://t.co/inGdDkb6PYfrozen11d ago100K2832734470.41%
General question: does anyone actually use OpenAI Dot, or pay for Codex just for Dot?frozen12d ago96K26611094130.41%
Fun fact: GPT 6 Luna Max feels unlimited. https://t.co/p5mCdQdONTfrozen29d ago72K9937601751.58%
We just one shotted Stardew Valley in 54 hours with Opus 5.5 It's crazy to even say it but it has its own storyline, every single feature we could imagine, including, crafting, planting, fishing, even the vibe itself. 12 agents ran in parallel the whole time, one per area of the game. each one builds, gets torn apart by a critic. I can say Opus 5.5 is the true goat. IMO It can truly surpass Stardew Valley if the loop finishes and even publish into steam right the way as demo. By the way we only spent $200 on this. The original creator spent years to get to this point.frozen28d ago68K464225873361.31%
That’s why I like X so much https://t.co/gmLCDYqdzSfrozen10d ago53K7604191211.52%
I made something crazy with Opus 5.5. Can't believe Opus is this good. It just mopped the floor with Astra.frozen29d ago51K5208100.12%
I gave Claude Fable 5.5 one challenge "Show civilization from day one to now. Go all out." No footage. No stock. Every frame is code. https://t.co/URA4st1yPHfrozen20d ago42K421242932241.66%
Tibo just pressed the Codex reset button. I'm not 100% sure why it has to be this time of the day, but my weekly reset was literally an hour ago. That's disturbing. @thsottiaux it's better to give us a bank reset, so we can use it when we need it.frozen26d ago39K4555564111.36%
A note for Claude: Claude Code right now is perfect! Stop adding features, adjustments or optimizations to Claude Code. Don't nerf, don't quantize, just leave it as it is until Opus 6 please! Thank you!! @claudeai @ClaudeDevs @bchernyfrozen27d ago29K1.1K35433184.00%
Sonnet 5.5 might drop today at 11 PT. Keep in mind Sonnet is already half the price of Opus. $2 in and $10 out against Opus 5.5 at $4 and $20. Anthropic said Sonnet 5.5 gets many of the same upgrades, and Opus 5.5 landed 40% cheaper per task than Opus 5. So we might get our first unlimited model if Sonnet drops another 40%.frozen24d ago25K2104120160.97%
Has anyone else noticed that OpenAI seems to be going downhill since DevDay?frozen8d ago24K53312602102.58%
Hahahaha OpenAI is cooked.. OpenAI launched Dots at Dev Day. https://t.co/xX98OYDMbk goes directly into Grok. Free ad for Elon Every promotion they make is directly getting into SpaceX. Elon just wins ahead of time.frozen23d ago23K25017050.21%
IDK today’s update is relevant for most codex users. Instant Steer is nice, but it already felt pretty fast IMO And Ultrafast on Sol? 90% of people here can't even use it Let’s make a reset pool? https://t.co/tTDcz36qeHfrozen14d ago23K169319120.85%
Founders, which model you like the most? 1. Claude Haiku 5.5 2. GPT Luna 6.0 3. DeepSeek v4.1 flashfrozen14d ago20K65140160.57%
OpenAI employee who got arrested. I’m guessing he got paid 100 times OpenAI salary for doing it. https://t.co/hiEh19TjSOfrozen9d ago19K1163210110.78%
Claude's new water wave design is so cool!! Also couple updates, including the new sessions load and the multi sessions UI. I started to feel like claude is the one making 28 days challenge instead of Codex. https://t.co/GfDpaovqpZfrozen8d ago18K2035131231.35%
Claude just did all of us in open source a favor!! OSS Scanner is free. Established projects apply, then Claude scans their code for security holes and emails the maintainer a suggested fix. Anthropic says brand new projects usually don't get in. I might going to opensource serein browser and try to apply this week. Let's see if that counts as established.3d ago16K122481471.14%
Someone build spiders man 2 wth AI before GTA6frozen26d ago15K51251100.46%
Claude just made a motion graphics video for the Stardew valley we created. We officially enter a new era of AI. https://t.co/aQMlVPz0aKfrozen26d ago14K73192320.82%
Fable 5.1 is routing to Fable 5.5 in the backend!! So the name says 5.1, but 5.5 is the one answering. I'm going to test a couple prompts. Reply with any prompt you want me to run. Can't wait to see how good it does!! https://t.co/xxukduVC68frozen21d ago14K753100120.72%
Tibo just cleared things out. Codex's $200 plan is 10x weekly now instead of 20x. That's the same $20 per 1x as the Plus plan. The bulk discount is gone. Claude's top plan is the same 10x on weekly. Basically OpenAI is matching with Anthropicfrozen23d ago14K77022120.75%
Long story short: we’ve got more usage now. Best update ever! Now we need new model like Astra 6.1 or Sol 6.5frozen6d ago14K65021060.68%
Muse wonder Full MAX is benchmaxxed. No word just watch how it compare to SWE 2 MAX https://t.co/7Fn7Lf5HNLfrozen10d ago13K782202150.90%
Don't think GPT Sol 6.1 is smarter than Opus 5.5. But it is faster, and it costs way less. https://t.co/iY9YUIKPlafrozen23d ago12K1756150131.72%
Hot take: use Claude Opus 5.5 xhighfrozen28d ago12K63340110.68%
As AI users, what's your favorite AI model this week? 1. Grok 4.7 2. Opus 5.5 3. GPT 6 Luna 4. GPT 6 Solfrozen29d ago12K41052110.81%
I just killed all browsers. Introducing Serein, with built-in Claude Code, Codex, and Grok agents that can browse the web. Each agent gets its own tabs and cookies, when they reach a password or payment page, they stop and hand the tab back to you. Serein works with modern agent harnesses like Claude Code, Codex, and OpenCode, using your existing subscription. Currently is free on mac, link in the comment 👇frozen9d ago12K717286371.29%
I can't imagine myself using Ultrafast with GPT Luna max. It's basically a premium, intelligent model at 750 tokens per second. Ultrafast comes with the $500 plan. 10 years of that is $60,000, about $16 a day. To own the closest open model, Kimi K3 (610 GB of memory): - 5 DGX Sparks is about $23,500. - 2 Mac Studio M5 Ultras is about $36,600. Neither gets near 300 tokens per second. So we're paying for OpenAI just to use this beast, in my opinion.frozen23d ago11K35012060.47%
NOOOO WAY!! Claude will always pick 17??? Ask it to "pick a number between 1 and 30" and watch. Try it out let me know! https://t.co/GOwDUDuTbjfrozen18d ago10K34021320.57%
This level of AAA game is going to be accessible to everyone across the world. A whole open world city that opens in a browser tab. https://t.co/WYcKevMDLJfrozen26d ago9.9K1067153241.57%
Gemini 4 Argon tomorrow! Which one are u guys excited the most? - Gemini Argon - Muse Wonderfrozen10d ago9.8K1795170112.16%
The new claude code app is so good i had to update Serein right away. For ppl who doesn’t know serein it’s my own browser, with claude code and codex built in. This is what it looks like now 👇 https://t.co/7e3zsVkGnWfrozen8d ago9.6K45214260.72%
Grok Bot is going to use Claude Opus 5.5 as a backend model!! Midjourney for images and Suno for music too. SpaceX says it'll use whichever model gives the best result for the task, so ElevenLabs for voice is probably next. This is going to be insane for grok bot users.frozen16d ago9.5K3332150.46%
X is introducing one plan for X, Grok, Cursor and Grok Bot! There's no pricing yet. Right now that stack is three separate bills at list. - Premium+ is $40 - SuperGrok is $30 - Cursor Pro is $20 Get ready for this one. It'll 100% benefit all of us.frozen21d ago9.4K4549080.70%
This new Claude Code feature hit me hard! Projects automatically spawns all your subagents in the cloud, and it can spawn them locally as well. Downside is it burns plan usage faster than a normal session, but we’ll get our tasks done like 100x faster and still hit the goal. BTW it hasn’t rolled out to my account yet, so I just signed up for early access to see what’s up.frozen13d ago9.1K582130311.14%
Claude just gave my account up to $100 of initial usage that doesn’t count toward my limits. It only works in Projects and expires 24 hours after setup. You can easily claim it by trying the new project feature on the sidebar. I’m going to see how fast we can burn the entire thing.frozen12d ago8.2K37016190.77%
Guys, check your Claude API credits expiry date. Mine expire in 4 days and none of it will rolls over, new ones come next billing cycle. I’m about to go burrrr https://t.co/2YXU3Q9Jd1frozen14d ago8.1K1103100.18%
I made a 25 minute animated episode with Claude, no AI video generator. I gave it the idea for the show and for every episode. Claude directed everything else and wrote the whole animation engine in code. Some props it built in Blender through Blender MCP. One of the main characters is a Claude Code terminal turned into a robot that keeps asking permission. The inventor hits yes every time without reading. (WHICH IS ME) It had rendered over 37,000 frames and kept redoing shots on its own until the scenes worked. The actual prompt is in the first reply.frozen5d ago7.9K773231662.14%
It’s 100% a troll. Just letting you guys know the owner of https://t.co/gkKehaEDQU is @StefanoGPT Pretty funny guy ain’t gonna lie.2d ago7.7K1402100.22%
Step 5 Preview just built an insane 3D landing page!! Ran it on OpenCode’s free week. All I asked for was a 3D landing page, and told it to go all out. It made a sculpture that actually orbits when you drag it, with almost 9,000 particles chasing your cursor. Didn’t think a preview model could do this. If you’re on OpenCode, Cline, Kilo Code or Hermes Agent, it’s already in there. You just switch the model. Nice job from @StepFun_ai here.frozen14d ago7.7K1928040.43%
Claude just coded out this entire 25 minute cartoon, 0 AI generated video!! The episode 2 of Gus & Spark is now out, the cartoon we're making with Claude. I'm calling this one the flying car episode, and it's actually fun to watch even though the quality still needs a lot of enhancement. It's a big one, Opus 5.5 worked on it for over 20 hours. - about 50K new lines of code - 5.75B tokens, about $1,670 at API prices - added AI voices this time (a local voice model), but every frame is still code - since it burns a lot of usage, we might just make it a weekly episode It has several easter eggs in it. Also just letting you guys know a certain someone is in there with muse. One of my dreams was always to make our own cartoon haha, and AI made it possible. The future is going to be so insane!frozen5d ago6.5K59591301.60%
Bad news everyone! The old days are gone. Tibo is not giving us free resets anymore. OpenAI just cut our usage to half. Seems like we have to all in dot insteadfrozen21d ago6.2K5058051.10%
Not sure how many people got into the infinite money glitch. Seems like it already sold out within 12 hours. No more freebies. https://t.co/bFEVNVIa1cfrozen17d ago6.2K2708020.60%
Meta is definitely on steroid. Wonder Full just scored higher than Gemini Argon and Fable 5.1 in Artificial Analysis. It just made Gemini Argon looked like a crown. https://t.co/J1qeHb2sU4frozen10d ago6.1K1506130.41%
Quick question: after a month of trying Astra. Is this model actually AGI?frozen18d ago5.9K28129000.98%
20+ OpenAI launches tomorrow? Here’s what I’ve heard so far: - Personal assistant called “o” - New Pro Max plan - GPT Astra Mini - maybe GPT‑6.1? - or personal assistant marketplace? What do you think the other launches could be?frozen24d ago5.7K29221000.92%
OpenAI is better than Claude for average people. 
Claude is better than OpenAI for builders. I know so many people who only know ChatGPT IRL. That's not a gap in quality. That's a gap in distributionfrozen22d ago5.5K87521132.11%
3 new OpenAI model names spotted: aurora-blue-alpha aurora-green-alpha aurora-red-alpha I’m guessing these are going after Opus 5.5, Sonnet 5.5 and Haiku 5.5. Fable 5.5 might drop next week too. Maybe OpenAI wants to see how their models stack up before deciding what to release. Seems like they have abandoned the Sol, Terra and Luna familyfrozen5d ago5.5K69216061.70%
Almost 24 hours of silence from OpenAI. No updates today. Get ready, a reset might be coming.frozen13d ago5.4K62310031.44%
Claude just confirmed fast mode in Claude Code still bills extra usage credits even when your sub still has usage left. I wish max users can use it within the subs.frozen7d ago5.3K6306031.36%
ChatGPT Pro Max showed up in the code at $500 a month. $200 of ChatGPT Pro lasts me about a day in Codex right now. At that rate $500 is about three days. Maybe it's better on Claude Max?frozen28d ago5.3K31113020.89%
Me at 21 Still remember my dog and Covid time.. https://t.co/x8UmFM6XCQfrozen23d ago5.2K1506100.42%
Space Bunny is now wired into Codex and Claude code CLI!! It's a new stealth model, nobody knows who made it yet, and it's free for the whole week. 1M context and it reads images too. I ran it on a real Codex Session today, 95K tokens so far, and it cost me $0. Let's build something fun with it. By the way what model do you guys think it is? hopefully not another router.frozen28d ago5.0K1816060.62%
IMO, Codex improvements is pretty simple: - 2× more usage on the 20× plan, or make Astra 2x more efficient - A model that beats Opus 5.5. Surprise us. - ChatGPT + Codex + dots in ONE system, with shared memory and context More usage. More capability. Less switching. Which one can we lands first?frozen18d ago4.8K69611051.90%
Claude Code and Codex can reverse engineer any apps now. It’s called REA, It’s an absolutely crazy skill to have. 1. Build the plugins an app never gave you 2. Pull your data out of apps with no export 3. Read how any feature you like was built 4. Build any cheats, scripts you want to have We got a lot of our stuff built this way. Repo below:frozen4d ago4.5K311130331.72%
Tibo just put Codex on a 28 day challenge!! Every single day it either ships one clear improvement for most Codex users, or does a full reset. This is going to be a fun month for Codex, start burning everyone.frozen18d ago4.5K4428111.24%
this is insane...frozen22d ago4.4K25030120.91%
wait… i thought we could always overwrite it 😭 Tibo rushing the day like he’s going to touch some grass https://t.co/IC4uABoPOTfrozen11d ago4.3K2704060.85%
Just got an email from OpenAI. They added around 62,500 credits to my account!! Seems like we can use them directly after we hit the limit in Codex. Just a reminder it will expire at Dec, 31 2026 https://t.co/0FLIb0Po7Efrozen23d ago4.3K43024001.58%
OPENAI IS COOKED.3d ago4.0K3911011.05%
Opus 5.5 just made a full heist trailer for our open world!! It's called "Nobody's Driving". A wheelman, the brains and a sea lion in sunglasses steal every self-driving robotaxi in San Francisco, in 99 seconds. - story shots are Higgsfield (Nano Banana Pro + Kling 3.0) - the heist map is built in Blender off our own city data, all 3,305 real buildings - it made 8 phonk tracks and kept the one whose energy matched the edit best - then a second agent ripped the first cut apart and it rebuilt the whole thing It started from this prompt: "make a dynamic <time> motion graphics video about <concept> that shows what an incredible motion designer you are, like it's your showreel for a résumé. go all out. Here are some references: <paste them if you have any>" Can't wait to make this one playable.frozen24d ago4.0K28510151.24%
Just got my second Claude Max yesterday. Claude's reset bank and usage is so much better than Codex's. Spend one and it doesn't reset your timer. Just your usage. Codex still pushes your reset date forward when you use theirs while their usage just gone in 1 day. https://t.co/3WzSPuAdM9frozen25d ago4.0K43311041.54%
Besides Theo, anyone else think Muse Wonder Full is S tier? Worth $50 just to try?frozen9d ago4.0K28020011.24%
Is it just me, or did Opus 5.5 getting worse those couple week? - Weaker answers - Hitting limits way faster Last time this happened, a new model dropped days later. Fable 5.5 incoming?2d ago3.9K99218043.18%
What about no 5h limits, same usage but release Astra 6.1 and make it 50% cheaper? Remember to do your vote. It decides what happens to our codex usage.frozen12d ago3.8K4208021.35%
Opus 5.5 &gt; GPT Astra better outcome and more usagefrozen1mo ago3.8K51113001.70%
GPT-6.1 Sol is out just like what we have talked about!! Keep in mind OpenAI says GPT-6.1 Sol is almost Astra level at 1/5 the price. Price per million tokens, in / out: Astra: $10 / $50 Opus 5.5: $4 / $20 Sonnet 5.5: $2 / $10 GPT-6.1 Sol: $2 / $10 Sol's cached input is $0.10, Sonnet's is $0.20. Cheap models are getting so good!!frozen23d ago3.8K2905020.95%
Don't see GPT-6.1 Sol in your Codex model list? Your Codex is just out of date. Sol 6.1 needs version 0.153.0 or newer. Update to the latest (0.159.2 today) and it shows up. Just tested it. Performance wise it's acceptable, but it feels like it doesn't even consume any usage https://t.co/Y0rymBCxWRfrozen22d ago3.8K3206111.06%
It's terrifying. I didn't even notice it was an AI coworker initially. what this opens up is: - your co-worker might be AI, joining a Zoom meeting with you - your co-founder might be AI, talking and working with you - your bf/gf could be AI as well The future is going to be so wild.frozen21d ago3.7K3215071.21%
Gemini 4 Argon is finally out after months of waiting! I can't believe it can create this quality of the 3D asset, with multi agent working on it. This quality of output is crazy. Hopefully it's not benchmaxxed https://t.co/Z4wacFM348frozen9d ago3.6K39460121.69%
Introduce Multi Claude lets you run multiple claude desktop apps on MACOS. I have 3 open on my mac right now and each has its own name and colored dock icon. Open multi claude tray and click "add account" and sign in separately in each claude window. Repo in the comments https://t.co/dzVu2rpCF7frozen11d ago3.5K371162332.51%
Another release from OpenAI?frozen19d ago3.5K4919011.70%
I'm not joking but OpenAI Dot is actually good at 3D designs. He made this on the first try. https://t.co/W90GvwOLbxfrozen11d ago3.5K60020182.57%
We might get Astra-minor model on Tuesday!! Remember to burn all your usage before then. Very likely we'll get additional resets.frozen25d ago3.5K3104001.01%
I can't believe Haiku is this cheap.. Been running Opus and spawning Haiku as subagents. It literally only used 10% of my weekly limit for the entire day. Opus alone would have used 20% on a similar day. So I just made it into a skill that we don't have to spawn it manually. - Haiku 5.5 for speed - Sonnet 5.5 for well-scoped work - Opus 5.5 for everyday intelligence Link in the comment 👇frozen14d ago3.4K464110162.28%
OpenAI Dot is actually pretty good!! It connected my ChatGPT and Codex history right after I jumped into Dot. It already knew me well, understood what we were working on and what task I wanted to do, and I hadn't said anything. It also created a little creature for itself!! Really impressed so far. It jumped into production right away. Now I will stress test it to see the usage issues.frozen21d ago3.3K2309131.09%
Space Bunny is quietly cooking on 3D assets. Hooked it into Codex, ran a single /goal and came back to this. Honestly might be the best stealth model out right now for frontend. Actually curious to know what's actually under the hood. Any guesses? https://t.co/IsuaX9uQyVfrozen28d ago3.3K51212092.24%
I haven't worried about usage issues since using Claude. On Codex I still have to worry about what I send over.frozen25d ago3.3K56010022.09%
I love the new Claude cloud sessions!! We don't need a VPS anymore. But it cost me over $14 on 1 prompt. All I asked was to check all PR in the repo. https://t.co/sJXnQjeqbrfrozen29d ago3.2K1706030.80%
Anthropic’s Fable 5.5 just spotted on Azure Very likely we will see it this week. Since The 5.5 series has been so good. Can’t wait to see what Fable brings.frozen9d ago3.1K5315011.92%
Hot take: Opus 5.5 is the best model in the world right now, period. Every week I max out my Claude before using Codex. I still remember it was reversed a month ago, before Opus 5.5 dropped.frozen6d ago3.1K51013022.14%
Claude Haiku 5.5 cooks even better than GPT-6 Luna!! It beats Luna on all 6 rows where Luna has a score. Terminal-Bench 39.2 vs 16.4. Computer use 72.4% vs 48.9%. Anthropic says it costs around 75% less than Haiku 4.5. Small models are cooking right now!! https://t.co/mjsXDYU7o5frozen15d ago3.1K1603120.72%
Being locked into Claude didn’t even realize Codex is having an outage. By the way codex has a new reset incoming!frozen27d ago3.1K2200000.72%
Respect to Alex Wang. He knows it isn’t better than Opus at coding, and he’s still actively replying to feedback and explaining why the outputs is bad.frozen10d ago3.0K1901020.72%
Codex Pro $200 is getting half of the usage. Tibo's words: it nets out at half the dollar in API spend compared to the old $200 plan. Same price, half the usage. Most of us already have a usage issue where it maxes out in 1 day. Now it seems like we will get maxed out in half a day. If openAI really want to make this work, show us how you can beat Claude in dollar per value.frozen24d ago3.0K3505011.37%
You can simply create any sort of motion graphics with Opus now. All you need to do is prompt like this: "make a dynamic <time> motion graphics video about <concept> that shows what an incredible motion designer you are, like it's your showreel for a résumé. go all out. Here are some references: <paste them if you have any>"frozen25d ago3.0K21010101.07%
https://t.co/LrODAO63Ogfrozen16d ago3.0K1206130.75%
Claude Code tip to get more work out of your usage. I have spawned over 120 subagents in 1 prompt. If you don't pick a model, your subagents run on whatever you're on. Start on Opus 5.5 and every worker it spawns burns Opus usage too. Say it in the prompt: "spawn Sonnet subagents or Opus, depending on the task." This way, Claude decides when to use Sonnet and when to use Opus. Same plan, a lot more work done.frozen20d ago2.9K2419061.37%
Meta's Muse Wonder pricing is out too, and it's matching opus 5.5 price with fable level model. Price per million tokens, in / out: - Muse Wonder: $4 / $20 - Opus 5.5: $4 / $20 - GPT-6.1 Sol: $2 / $10 - Fable 5.1: $10 / $50 - Astra: $10 / $50 BTW Wonder Full is only on the Muse Code plans.frozen10d ago2.8K1703030.82%
Haiku 5.5 is cooking hard! It produced the same level as Luna 6 IMO. It looks good and it's very fast. Same Blender motorcycle, same prompt for Luna 6, Sonnet 5.5 and Haiku 5.5!! "Make motorcycle with better materials in blender." Haiku 5.5: 40K triangles, 12 mins Luna 6: 123K triangles, 40 mins Sonnet 5.5: 175K triangles, 44 materials, 35 min What do you guys think?frozen15d ago2.8K2914131.38%
Everyone had used ChatGPT’s new UI answers, but this guy actually built his own and open sourced it. OpenUI 1.0 is MIT licensed. They even rebuilt ChatGPT’s Intelligent UI on top of it as a repo you can fork, and it runs on local models through Ollama. If you build with Codex or Claude Code, one command teaches your agent how to use it: npx @openuidev/cli@latest setup-agent Their benchmark also has it at 67% fewer tokens than the JSON way, so it’s lighter on limits too. Nice one from OpenUI here. Opensource must winfrozen7d ago2.7K27171171.94%
Codex Auto-review is now free for everyone in codex!! It's under the "Approve for me" tag. A second agent reviews every action the main one takes, so you stop approving everything by hand. Claude users already know this idea. Now Codex has it too. Day 2 is acceptable but I feel like we need additional reset here. Let's see what day 3 hasfrozen16d ago2.7K802010.40%
Anthropic just won. Opus 5.5 is so good it can generate any type of vehicles in high quality. https://t.co/e8Cxh8nD9sfrozen29d ago2.7K27211041.63%
Can't wait to see Fable 5.5!! It already showed up in Azure's model list last week as claude-fable-5-5. One of the biggest downsides I have right now is I can't identify all the issues in my code. If Fable 5.5 can find them all it will be wild!!3d ago2.7K3046031.61%
Claude keeps giving us more, and Codex keeps taking away. Max 5x is $100. Max 20x is $200. Both now come with free API credit, so we get to use extra Claude models for free. Meanwhile Codex cut our limits by 50%. Free credit on top of the sub. Let's gooo. https://t.co/0TqigZy6Egfrozen15d ago2.6K2722121.32%
I've seen a lot of AI trading posts. Most are hype. This Kimi K3 + GPT-6 Astra one is different. It explains why a stock moved, with a source on every claim, and never tells you to buy. The best part is the three agents. One builds the bull case, one builds the bear case, and a judge checks every claim from both against the data. And it's cheap. Running Astra on everything would cost $60,000 a day. Filtering first and letting Kimi K3 do the reading gets it to about $450. simply copy the post and give it to your codex.frozen26d ago2.6K14020141.17%
@thsottiaux do you mind telling us if the next Codex reset is a bank reset or a regular reset? If it's a regular one, when is it coming? It's getting intense now.frozen27d ago2.5K1705000.87%
Finally we got more compute for OpenAI!! I’d say it’s a pretty good update. I hate seeing all those “selected model is at capacity” errors while it’s running. For us it should mean slowly getting more crazy models like GPT-6 Astra, just more efficient. IMO very soon we will get more usage from OpenAIfrozen7d ago2.5K2504021.23%
Watching all those things made me realize. The best job in the future is IP lawyer.frozen19d ago2.5K29011051.81%
The ultimate AI stack right now: Grok Bot logged into your Claude Code. Every Grok Bot comes with its own computer, and it signs into tools the way you do. So it can set up Claude Code CLI on that machine, start the login, and send you the auth link to approve. Codex works the same way. Then one instruction does the rest: "run /loop on [the task] and message me every time something ships or breaks" /loop is Claude Code re-running a task on its own. Give it /loop 30m and it wakes every 30 minutes, or plain /loop and it paces itself, building while your laptop is closed. Every Grok bot has its own machine, so a second bot is a second loop running somewhere else. No more VPS and you can use it everywhere. Claude's own cloud sessions charged me over $14 for one prompt, and this runs on subs you're already paying for. You can open that computer from your phone and watch it work from anywhere.frozen29d ago2.5K17040131.37%
OpenAI is cooked. Most of us are picking Opus 5.5 over 6.1 Sol It’s way more value, faster and it feels 10x better. @thsottiaux can you make the come back in 28 days? Ship something smarter and more efficient than Claude? https://t.co/CrnbGdCIA7frozen17d ago2.5K35213042.17%
Run /checkup prompt-audit in Claude Code now! It checks your CLAUDE.md, skills and agents for stale stuff. Just ran it myself and removed 20+ skills and MCPs that had been idle for days!! https://t.co/OReGK8Jr2Bfrozen27d ago2.4K1619041.23%
Tuesday -&gt; Gemini Argon??frozen10d ago2.4K4218022.23%
Seem like the new aurora model from OpenAI is the image modelsfrozen4d ago2.3K1105000.69%
Elon just confirmed SpaceXAI is getting renamed to SpaceXSI. It's a very big opportunity for everyone to get into .si domains right now.frozen18d ago2.3K3415121.86%
Just a reminder for all Claude users Claude subs doesn’t cover Fast modefrozen13d ago2.3K2004031.19%
Sonnet 5.5 just made Plants vs Zombies in 2.5D!! It looks higher quality than the original. It also added our own custom features, like drawing cards and abilities. This is actually something I'd like to play. https://t.co/q4ZGdf8v3Nfrozen22d ago2.3K2804091.81%
60% less disk space in Codex? that's huge for me. My macbook pro is always full of docs and code. I have to clear space every week just to keep working. my weekly cleanup prompt below https://t.co/o4iRBPGECrfrozen8d ago2.2K19015031.69%
Just a reminder for all codex users. You can now directly use codex usage in devin https://t.co/A8xAGg28qGfrozen23d ago2.2K902050.73%
If you want to build this level of 3D motion graph website. Like the agentic harness website I'm about to release, fun 3d with scrolling, based on your product. Here is the prompt: "Build me a high quality 3d motion graph website for serein where we can linked to the vercel https://t.co/X7SBufnpNs web page and give a wow feeling with high quality 3d motion graph website for serein, go all out" Here is the result below.9d ago2.2K28050142.15%
Everyone's hyped on OpenRouter's new Jev router. Theo just stress tested it. Be honest I Love Jev. But no reasoning means it can’t actually act as a main route. You're paying more, waiting longer, and getting Astra-low-effort at the end of it.frozen26d ago2.1K1304020.89%
Founders, you can build literally anything you want with Opus 5.5 now. No excuses left!frozen19d ago2.1K3018101.89%
Codex just increased the default speed for both GPT-6 Astra and GPT-6.1 Sol. From 30 tokens per second to 50. That's a huge update! It used to be slow as a snail, so we had to focus on fast mode. Now we can use default mode and save some usage to run more 6.1 Sol. It's the Day 1 update, now let's make a smarter model than Opus 5.5!!frozen17d ago2.1K1915001.20%
"Selected model is at capacity, please try a different model." This is the one we really need fixed!! So many people are hitting it, me included. I trust OpenAI can make the comeback, but first we need to have a steady model. https://t.co/BDTQsXnuMD2d ago2.0K20211021.72%
How tf Claude is getting so good now! Sonnet 5.5 is a big step up in agentic work, 30% cheaper and faster than the sonnet 5? Anthropic might be back on track. We just got ourselves infinite premium AI models.frozen24d ago2.0K1416031.19%
Every founder, this is how you do distribution maxxxxing in 2026. Over 20M+ views across six videos. One post alone did 7.5M and brought in 3.3K followers. If you're not sure how to make money from that or connect it to your product: After you get the views, people follow your account. Companies will want to partner with you, or you can sneak your product into the video. Pedro does the second one. His quit-drinking app just shows up as a lock screen notification in the post: "12 days and $160 you haven't drunk." (The sneaky way) He still picks every idea and approves every post himself by the way!frozen28d ago2.0K13130151.60%
If you pay for a Max plan, you're already in the 2%. And if agents run loops for you 24/7, you're 10 years early. We've had ours running since June. One /loop ran for weeks over 80 sub agents. It's still early!! Imagine 10 years from now, people want to get into AI and we have 10 years of experience.frozen19d ago2.0K1516011.16%
Karpathy has a simple idea for understanding what AI gives you back. AI will do more and more of the work. So our job moves up to checking it and understanding it. The easier the output is to read, the better. He ranks the formats, and each one beats the one before it: - Text: "explain this in ASD-STE100" (the strict, simple English airplane repair manuals use: short sentences, one idea each) - Diagram: "draw this as a diagram" so you can see how the parts connect - Web page: "give me this in HTML" and you get a page you can click through - Video: "make a 3Blue1Brown-style explainer" (the animated math YouTube look), with a voice Step 4 is the one he's most excited about, so we made one from his own post. 1:44, narrated, captioned 👇 We also packaged all 4 steps as a free Claude skill that asks which level you want. Links in the comment.frozen19d ago2.0K2116081.84%
Hopefully Gemini 4 Argon isn't benchmaxxed. It landed at #8 on coding but #1 on talking looks like it’s a solid orchestrator brain. just not a better coder than Opus 5.5.frozen21d ago2.0K2327011.68%
EU friends, OpenAI now has a watermark for text too. I guess it's one of the daily improvements. It's invisible, and it's coming to ChatGPT and Codex output in the EU over the next few weeks.frozen17d ago1.9K1104020.89%
Wait. Anthropic just dropped Claude Duet, its own version of Dot / Muse / Grok Bot!! Two weeks ago I posted that we don't need Claude to drop its own personal assistant, we can just use Grok Bot. Duet already has your Claude memory and apps. Grok Bot seems to only send the complex stuff to Opus 5.5. So why should I use Grok Bot if Duet is native and runs Opus 5.5? Just enrolled into wait list to see what’s up!23h ago1.9K28113092.67%
Thanks for the reset today Tibo Since we already got one yesterday, this one's basically banked. love that you thought of that So for day 4... can we get Astra 6.1 instead!frozen15d ago1.9K2214011.51%
Claude Code has a new feature called mods A mod is a small plugin that adds your own UI and rules inside Claude Code. It reloads live while you edit it, so you see every change in the same session. - A side panel with live data - A chart right above where you type - A line in the status bar, or a pop-up notification - A rule that checks every command before Claude runs it, and blocks the ones you never want, like rm -rf - Your own slash command To make one, tell Claude Code "make a mod for Claude Code that..." and describe it. It writes 3 files: .claude-plugin/plugin.json hooks/hooks.json hooks/register.tsx Mine puts my Codex and Claude Code requests in one live chart above the prompt. The easiest one to start with is a status line. It's a single line of code.frozen18d ago1.8K20110031.84%
Claude is down.. Did openai just hacked claude? https://t.co/SMkCecUiP0frozen23d ago1.8K1805001.26%
You guys are going hard on Tibo. I have never seen a post get 2 community notes at the same time. https://t.co/IWlvmaFrGUfrozen12d ago1.8K1603011.10%
Meta might actually be back in the AI race. Putting Muse directly into AR glasses is an absolute banger. This is what AI hardware was supposed to feel like IMO.frozen29d ago1.8K702000.50%
I guess day 9 update is a reset as well? I'm ready to burn some tokens9d ago1.8K3315002.16%
Can anyone explain to me why Dario looks so upset? https://t.co/niauJIwcZ7frozen21d ago1.8K1309101.28%
GPT-Sol-6.1 cooked. Huge congrats to the team. $0.10 input cache and topping Opus 5.5 on Terminal Bench 4.0 is not a small thing. The just bar moved again. Now let’s test it in real world workflow.frozen23d ago1.8K1201030.90%
We got CJ raising startup fund before GTA62d ago1.7K2316021.89%
Codex right now. https://t.co/w6FcLLHffWfrozen14d ago1.7K19010011.77%
Unless something bad already happened, I don't think they will be talking like this. I feel like a catastrophic AI event is happening right now, and they'll make it public in '27. We just don't know yetfrozen13d ago1.7K1002010.77%
We're about to achieve superintelligence and still can't figure out if it's Unites States or United States. https://t.co/M70SXtBPHgfrozen22d ago1.6K1407001.28%
OpenCode Console got breached. If you saved your OpenAI, Claude or Google API keys in there, rotate them now. Revoke them on each provider’s own dashboard. Swapping the key inside Console leaves the old one alive. Local harness should be fine.frozen5d ago1.6K1203010.98%
That's a good start from Anthropic. But public good is hard to define. Maybe just give the shares to the entire public, universal basic shares (UBS).frozen4d ago1.6K1203021.05%
Just shipped new versions of Serein. Added wallpaper support, so you can switch between different themes. Next up: we’re planning to make it easier to use, add the generic harness part, also the personal assistant + opensource. https://t.co/ypmszwhxsn17h ago1.6K24310062.68%
Sad News: Anthropic did not remove the Fable limits. it still there you just have to maximium it https://t.co/lkK6owQeLK1d ago1.6K2107021.87%
You can tell your agent to ramp up the run and drop to slow-mo on the jump. Tesseract just added optical flow and frame blending, so the slow part doesn't stutter. Doing speed ramps by just asking is crazy!!frozen17d ago1.6K601040.70%
We are at 300+ downloads in one day and many requests for windows version. Me and @Av1dlive are shipping windows very soon and going to prepare to opensource it as well. https://t.co/xv4gw5NMCNfrozen8d ago1.6K1526061.85%
Grok Bot is going to run on Opus 5.5!! Elon says it'll use the best model for each task, and Opus 5.5 is one of them. We don't need Claude to drop its own personal assistant anymore. We can use Grok Bot directly. Can't wait to put it to work.frozen15d ago1.6K1204021.15%
Arcads just launched Singing Ads. They say it's crushing on Meta, so here's the money side. 1. Find the client first. Dentists, gyms, restaurants, roofers, anyone with a clear offer. Search their city in the Meta Ad Library and pick the ones running a boring ad. 2. Paste the offer into GPT-6 Astra and ask for 5 four-line songs. Pain or offer first, then proof, then the business name as the chorus, then what to do next. 3. Keep each song near 15 seconds, hook in the first 3, and drop the lyrics into Arcads. 4. Make one ad image in ChatGPT Images 2.5 with the storefront and the offer on it. Use it on all 5 so only the song changes. 5. Put $5 a day on each for 3 days and keep the cheapest click. 6. Send the client the winner. Charge per winner, not per video. Save this one, the client search is the part most people skip.frozen22d ago1.5K1003000.84%
Tibo just pressed the Codex reset button! My usage hit 0 yesterday and I was thinking about using my reset. Glad I didn't. Go all out everyone, we're probably getting a reset in our bank on Sunday too. Thanks Tibo! https://t.co/2OI3kmLiTHfrozen12d ago1.5K1102010.93%
Imagine OpenAI drops Ophone todayfrozen23d ago1.5K1903011.54%
OpenAI has a wallet incoming on mobile. Which means soon your agent won't just book the flight it'll also pay for it. We're about to find out how much we actually trust these things. https://t.co/HKlCV3yHPofrozen21d ago1.5K1914011.67%
this is absolutely terrifying...frozen13d ago1.5K1202041.21%
Codex Router got an official website and I didn't even know until today. Last one to find out about my own picker's website. Link in the commentfrozen22d ago1.5K1206011.30%
NOOOO! I want Astra 6.1 not Terra 6.13d ago1.4K2704132.43%
Wasted! https://t.co/MTrPfWAF9Sfrozen28d ago1.4K1305001.26%
OH NO Guys, starting Nov 12th. Anthropic's new policy says no more abusing Claude. We CAN NOT ROAST CLAUDE ANYMORE if he made mistake.frozen14d ago1.4K815000.98%
Codex just reset our limits!! They shipped 4 things on the same day too. 1. Approve for me (auto-review) is now included. Their note says it can use 2-10% of your plan when it runs. 2. Simplified API for builders. 3. Meeting notes built in. 4. Decisions API is live for builders. What a day. Drop what you're cooking with it below Let's gooo!!frozen16d ago1.4K1707011.78%
Dot feels like an infinite Astra 6 glitch. I keep throwing work at it. It keeps shipping assets and code. The usage is wild. The quality makes it even crazier. If you haven’t tried dot yet, you’re missing out. https://t.co/rmAasE5dDWfrozen9d ago1.4K1808042.16%
This is how many Claude Max you need for the weekly limit: 1 Max if you just prompt while editing or run a 1 to 2 agent loop. 2 Max if you run a 1 to 3 agent loop with subagents. 3 Max if you run a 4 to 6 agent loop with subagents.frozen15d ago1.4K1606041.88%
Where is dots? https://t.co/YuPXtpMSfH2d ago1.4K1506011.61%
Google just dropped its own vibe gaming platform but it’s giving me a 403 error.. Has anyone tried it yet? https://t.co/Ee3CI8cB6Afrozen15d ago1.4K1107011.40%
Going to drop my first harness today! stay tuned guys! https://t.co/WLuurwKz96frozen9d ago1.3K1308021.74%
Gemini 4 Argon is coming with the greatest comeback!! On Google's own chart it beats GPT-6 Astra, Fable 5.1 and Opus 5.5 on 13 of 19 tests. - DeepSWE: Argon 77.9, Opus 5.5 74.2, Astra 74.1 - Reading a huge codebase (256K to 1M tokens): Argon 84.2, Astra 71.8, Opus 5.5 66.8 - It costs $2 in / $10 out per million tokens for now. Astra and Fable 5.1 are $10 / $50, so it's 5x cheaper. It's 2x cheaper than Opus 5.5, which is $4 / $20. Only trusted cyber defenders have it today, API "as soon as possible". Can't wait to test it in real life.frozen22d ago1.3K602000.61%
Huge congrats from @LLMJunky. Launching it’s own model is pretty dam cool3d ago1.3K804010.99%
Motion designers are now dead! Opus 5.5 + Mirelo is gonna be a crazy combo!! with mirelo sfx 2.0, we can add sound effects and ambience to motion graphics videos. it watches the video and syncs the sound to the motion. it also removes unwanted music while keeping the dialogue.frozen8d ago1.3K15010101.98%
Grok bot users, a domain can be a good investment if you get the right name, and .si is getting a lot of attention. Trump signed an order telling federal agencies to say SI instead of AI. A .si costs $25 and one sold for $20,000. Most .si flips never sell though, and the wrong name is $25 a year for nothing. We just made a Grok bot template in the comment that checks domains under $100 for AI or SI related names so you can get started!!frozen13d ago1.3K1313051.71%
9 of my jobs died halfway through last week. Out of usage, mid run in codex. I made a spreadsheet of every plan and when it resets, updated it twice, then forgot it existed. So I handed the job to Viktor, our AI employee, in Slack. 27 minutes later: my main plan is empty by Thursday every week and doesn't reset until Tuesday, and 7 of the 9 dead jobs were test loops that never needed the top model. He asked me before moving anything, and now he pings me at 80%, before it runs out. Try free at @viktor_com. $100 in credits, no card. Full link in my first reply.frozen27d ago1.3K805021.17%
Why does our open world NPC look like Slender Man?? 😭 she's supposed to be a normal girl walking down the street. Game devs, if your characters ever came out stretched like this, tell us what fixed it Until then she's the tallest guy in the city https://t.co/GgtzgqUZ7Ufrozen25d ago1.3K1115001.34%
No way, this dude just dropped a Discord version to manage your agents!! It’s a very interesting way to handle it not gonna lie. IMO I like Discord better than Slack.frozen14d ago1.3K1503051.84%
Remember to touch grass If you can’t, here’s a panda for you. Bro didn’t make a single sound https://t.co/qQpQO8Okt4frozen4d ago1.2K1204011.38%
NO WAY. Woke up today to see we have a flying car!!frozen6d ago1.2K2113022.21%
OpenAI launched Dots, but https://t.co/xX98OYDMbk was a clothing store. It sat there for years, and now the site just got updated. It’s kind of obvious.. https://t.co/xAAqxRkWhm3d ago1.2K1411011.39%
Grok Bot doesn't send every message to Opus 5.5. Seems like only the complex ones go to Opus 5.5 rest of it might still be grok.frozen15d ago1.2K1413111.68%
How to fix the PHP bug, the easy version. WordPress or a normal web host: message your host and ask if they've patched PHP2Shell. That's all you need to do. Own Ubuntu or Debian server, run sudo apt update && sudo apt upgrade then restart PHP with sudo systemctl restart php8.3-fpm (swap in your version, or restart apache2 if that's what runs it). Ubuntu 22.04's PHP 8.1 still gets fixed through apt. Any other 8.1 or older gets no patch, so you have to upgrade, and CISA says keep the site offline until you do.21h ago1.2K711010.85%
AI can now be our doctors and issue prescriptions!! Imagine our personal assistant, like Grok Bot/Hermes, connecting through Nolla and understanding what's happening with our health all the time. It could read our heartbeat from Apple products too. We will live long!!frozen17d ago1.2K1302001.27%
A .si domain sold for $20,000. They cost $25 to register. We ran a /goal in Codex to find 100 real-word ones. 4 were open at $25 on Oct 5!! https://t.co/6Ci4H4jFhK https://t.co/kDDWdFkPqB https://t.co/HeWRaVoyP6 https://t.co/Mv6nlQriHa The other 96 are in the article below.frozen16d ago1.2K1004021.36%
Imagine a world where Dot and Muse also uses Opus 5.5frozen15d ago1.2K1003001.11%
Here's how I let 50+ Claude subagents work on one repo for a month. 1. Every subagent gets its own worktree, a separate copy of the repo. Two agents never edit the same file at once. 2. Each worktree sits on its own branch, so nothing touches main until I say so. 3. When it's done, I merge everything into one trial branch and see what happens. Time to spawn multiple agents at the same time.frozen17d ago1.2K1208021.91%
Feel like the world is going to collapse.18h ago1.1K1802011.83%
Claude cooked so well with Opus 5.5, Sonnet 5.5 and Haiku 5.5!! Can you guys imagine how good the legendary Fable 5.5 is going to be? Can't wait for fable 5.5 next week!!frozen15d ago1.1K1713001.86%
Tibo just press the reset button!! What are you guys going to cook this week? https://t.co/C7nFlAS8Y1frozen20d ago1.1K903011.16%
Meta first fable level model is going to release today! Muse wonder 1.0 👀 It’s so silence didn’t even hear anything out of it. https://t.co/aiozOYTXy8frozen10d ago1.1K1800111.80%
The mystery model Le Chaton Fat is real! It's Mistral Large 4. Another open source model with over 1T parameters. Mistral says it beats GLM 5.2 on DeepSWE, a coding benchmark. 62%. Pretty good open source model from the EU. Keep going that way!! https://t.co/TtpDH3VZhifrozen16d ago1.1K1305021.89%
I doom scroll TikTok every day. The only thing that stops me is drama. And AI drama ads are crushing it everywhere! Pixar 3D, talking animals, claymation, realistic, any style. And you can create all of them on Arcads. A couple falling apart over a problem the product solves. An absurd mess where the product ends up the hero. The ad is the show. The first ad that stops my scroll is a win, and I’m buying whatever they’re selling!frozen14d ago1.1K903011.23%
This is how to automated the feedback loop for any product. Viktor reads our user feedback every hour. This is the message we sent him in Slack: “Every hour, check the feedback table in our database. Group new entries by issue. If something is broken, find the cause, fix it, and open a PR with the tests passing. Ask me before anything goes live. Next hour, check whether the same complaint came back.” First hour he found 1-3 new entries and some of it were the same bug Try free at @viktor_com . $100 in credits, no card. Full link in my first reply.frozen11d ago1.0K1204031.83%
lmaofrozen26d ago1.0K1100001.08%
Me and Claude have been cooking something. Anyone want to guess what it is? https://t.co/eKMYOUYik2frozen7d ago1.0K1507022.36%
"hyperscalers don't always have as many GPUs as enterprises require" So 323 places rent GPUs now, up from 209 in under a year, and the big clouds still run out. Good luck making 323 accounts to find one free H100. @shadeformai just puts 30+ of them in one console. Whatever the bubble talk says, nice one from Ed here.3d ago993901011.11%
Codex widgets on iphone are actually so good I can see my weekly limits right on my home screen now, plus whatever tasks are running on my mac it's under chatgpt in the widget gallery if you haven't added it yet Go try it out! https://t.co/kqCAIctQbRfrozen16d ago945805011.48%
WTF gemini 4 Carbon already?21h ago9411206012.02%
We're already at Opus 5.5 and we still can't fix the hands on any asset. It opens them so weird, like a monster. https://t.co/8wIsHLsR2L1d ago9301807022.90%
Just turned this AI UGC flow into a Grok Bot template you can run directly. - Apify pulls outlier videos - Grok breaks each one down: hook, tension, payoff - Test opening hooks on the same body so you see what moved the result - Higgsfield generates - Postiz drafts & schedules Template in the comment, full breakdown in the article below 👇15d ago8431003031.90%
Codex or Claude Code can build the whole app in an afternoon. Then the evening goes to servers, env vars and SSL. @synteq just shipped the evening part. Push to GitHub, one command, or drag and drop. They’re giving out $20K in credits for the launch. Comment LAUNCH on their post and grab some.2d ago798600051.38%
Anthropic just launched a startups program!! Eligible startups get a year of free Claude credits. Just applied myself to see if we can get into it. Let's see what happens!! https://t.co/qUcsNXZc53frozen16d ago784202000.51%
If you don't know what to run with Muse, Grok Bot or Dots. Here's one that took a fresh TikTok account to 1M views!! Pedro did it with Muse over WhatsApp. Brand new account, TikTok slideshows, 1.1M views and 2K+ shares in a few days. For research he sent it something like this: “ Find 10 posts already doing well in my niche or close to it. For each one, tell me the hook, the first slide, why people keep swiping, and how you'd use the idea for my app without copying it." Then he asked for 3 versions of that post and made the best 2. Use it to help you scale your own accounts2d ago717903042.23%
Every Dario post is important, he just replied to the guy who said anthropic quietly dropped ai safety. BTW it’s wall of text, feel like reading a book. The thing he keeps coming back to: you can’t pull the industry toward safety from behind. you have to be the one out front. so in his head, winning the race and doing safety are the same job. He’s got receipts for it, all of them stuff that cost anthropic money. - bio filters so strict biologists complain - holding cyber-capable models back for months - pushing for a tougher version of the sb53 safety bill - china chip controls - 5% of the company into a public benefit fund then he more or less shrugs and says people just count the bad stuff and i will probably ignore all that.1d ago713612011.40%
As a founder, your knowledge or experience shouldn't be the only thing that stops you anymore. i used to think oh i don't have experience, i don't know how to do xyz, i don't know how to do marketing. all of those are just excuses now. here's the whole stack for $0: 1. build it. download opencode, pick deepseek (free right now), and have it write your product. 2. distribute it. create UGC daily share it in tiktok, reels. 3. execute. that's the only part left that's still on you.21h ago708401010.85%
Founders, If you have an app and no content yet, this is the cheap version of his setup: 1. Find one video format that's already working in your niche. 2. Make it yourself, use the script, the assets, and that video as the example. (later you can hire more creators if you made the profit) 3. When a video hits, post it again on a second account with a clip from a big viral page up front. 4. Let Grok Bot schedule each morning's videos through Postiz to TikTok, Instagram and YouTube. Scale your product and cook your way there!frozen10d ago674702072.37%
Monday it was 323 places renting GPUs and the big clouds still running out. At the same time plenty of teams are sitting on GPUs they already paid for and barely use. @shadeformai Flex VMs lists those boxes with one curl command. They say teams on contracts can earn back up to 43% of a B300 deal by renting out the hours they weren’t using. Good to see the idle ones finally earning something.2d ago673800031.63%
Imagine robotics gets good enough to handle most production. What do humans do with all that capacity? The Interstellar idea here is to aim at goals bigger than what our machines can already achieve. Abundance should give us more to build, not less!!1d ago655902011.83%
the ai slideshow pages pulling millions of views all share one boring thing. the same face on every single slide. the catch is you can't prompt your way there. generate each slide fresh and the face drifts, and the page ends up looking like random junk no one saves. the ones that hold up train one character once, then feed that same one into every slide. it's an mcp call on a loop, same as what i run for code, just carrying a face forward instead of context. skip it and the page reads like a feed of strangers.1d ago545400010.92%