Pinned Tweet
Once a day I regret going the Investment Banking and PE route and not doing CS in college, but still taught myself code and still won a Google hackathon. Proud of myselfš¤ shout out @GoogleDeepMind @GoogleAIStudio shoutout @NanoBanana for being elite.
decided last minute to build for Googleās Nano Banana Hackathon. Had fun with this one. Not bad for 15hours of codingšāāļø. gemini is elite. found a simple architecture that allows for consistency & coherence over long stories. Just gotta add veo3 now kumayiri.com
Yāall were genuinely overly tripping. the 6.2 models are great, 6.2 Astra is an amazing model. the speed bump in itself is a huge improvement, while improving cost efficiency too. this model is the model the corporate streets needed for everyday work.
With the 6.2 Astra launch I think people are forgetting that as models get better and better the more you hand hold and the worse they can perform
Iāve been using GPT-6.2 Astra prior to launch & I found it very uneven & weird. On one hand, it was harder to steer & work with than other models. On the other, when given an ambitious project (make a space elevator ride from scratch with Blender & Unity) it could deliver well
1
29
Opus and Fable have been driving and creating some pretty neat marketing videos for my weekend projects. I hate that codex and astra canāt come up with output like this still, no matter how smart
4
70
How microsoft had not built this yet when thereās literally open source github codes for it is beyond me lol
Introducing Claude Meetings: add Claude to your teamās video call and it builds docs, whiteboards, and other artifacts with you while youāre still talking.
Starting to roll out in beta on Team and Enterprise plans.
2
65
I have a theory that the Jensen news over the past couple weeks was strategically timed and exposed by Anthropic so that everyone holding NVIDIA will rollover to Anthropic stock, and then theyāll soon start talking about how they are building chips internallyš¤
5
94
yeah theyāre setting my boy jensen up. releasing news like that during earnings season is a set upā ļø. and during front and center AI discussions and the political climate that comes with it. watch us find out that honey pot was actually U.S asset to set him upšš
1
2
79
Oh man. Hospitals and blood banks used to be the line even the worst hacker groups claimed they wouldn't cross. Now they're messing with the freezers. We are headed towards nothing good.
JUST IN: French blood bank hit by cyber attack with blood samples being heated to 110 degrees Celsius from their maintained -30, per Le Canard EnchaƮnƩ.
Readers added context they thought people might want to know
The 110 °C figure was the value hacked equipment was reprogrammed to reach, which does not mean the cold rooms or plasma bags actually reached 110 °C. France's health ministry said the anomalies were spotted in time, with no operational consequences. cyberattaque.org/cyberattaque-cā¦
54
Cheicolate retweeted
maybe referring to this article, for those interested already-clean.goodfire.com/
Yeah thereās no way labs donāt have RSI models internally, theyāre holding back some demons for sure
We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math.
It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x and even instruction following 3x, without gradients or training.
We call this Weight-Space Programming: frontier LLMs can now natively read/write neural network weights! š§µ
Blog: rikiyat.github.io/blog/weighā¦
Readers added context they thought people might want to know
LLM was given weights explicitly as text in prompt and used code/forward passes (per authors' logs) to ablate Qwen looping direction, a known thinking-mode issue per model card. Primarily fixed repetition in that mode, not general 'natively' improving weights/math.
rikiyat.github.io/blog/weight-spā¦
github.com/RikiyaT/lm_undā¦
huggingface.co/Qwen/Qwen3.5-0ā¦
huggingface.co/RikiyaT/Qwen3.ā¦
3
147
Cheicolate retweeted
Replying to @simpsoka
Jules should have been the lead from get go, not antigravity
i too got a plugin coming soon, @thsottiaux @jxnlco, can we expedite review process please š
Weāre excited to announce our new CapCut plugin is coming soon to @ChatGPT at CapCutWorldLA.
The plugin brings CapCut's video editing and generation tools alongside ChatGPT. You can start with an idea in ChatGPT, bring it to life with the CapCut plugin right inside ChatGPT, then take your draft video to CapCut's web editor to refine, share and publish.
By bringing the CapCut product you already love and use in ChatGPT, we're making content creation more seamless and efficient.
2
55
Not everyoneās cup of tea but Kimi 3.1 is good on my blockš«” use the API
GLM 5.3 flash is genuinely amazing, with computer use too, still thoughā¦thereās something about those kimi models that is cracked, they has a secret sauce, impressive new GLM and Deepeseek only make me more excited for that next Kimi drop
56
woahhh, this is new and very interesting. running on local off the shelf hardware too. well done @RagavanThurair and team, Dessa feels different. Havenāt felt a jump like this since sesameās beta launch
We believe voice is the most natural way to interact with intelligence, so we created Dessa, the most authentic full-duplex speech model ever built. Talking with it feels like talking with a personā¦full of emotion, nuance, pauses, and a natural cadence.
block.xyz/dessa
67
itās not a fair game š hackers/attackers only needs one opening while a defender has to keep track of whole systems. thereās genuinely no way to fully protect your systems without AI and 24/7 agents in this age of cybersecurity
I got hacked on multiple fronts this week. one of them appears to be Chinese but unclear. sht was and is still stressful having to go through all projects, patch them up, harden, and run security tests with as many edge cases, while also creating watchdog agents for monitoring. my roomateās school also got hacked with ssn numbers and other PII data breached. so once i started seeing the php and node vulnerabilities, it all started making sense. a new era for cybersecurity. buckle up.
55
they finna hack this just to make a point š
Today we announce 1Password for Agents platform, building with companies @Muse, SpaceX Super Intelligence Grok @Bot, @AnthropicAIās @Claudeai, @OpenAI, @TownAI, @Browserbase, @Browser_use, @heydana_ai, Instinct, @usekernel, @Perplexity_ai, @tryramp, @Tomo, and @wajo_ai, with more integrations on the way. š
We've heard from so many of our customers who want to use these new personal AI agents to help them get more done, but want more control over what the agent can access with the same trust and ease of use they get when using 1Password directly.
1Password for Agents gives your agents access without giving up your passwords. When an agent reaches a sign-in form, it requests a credential from your vault. If you approve the request, 1Password directly auto-fills the credential in the destination site so the agent canāt see or take possession of the credentials.
The agent gets the task done, but it doesnāt get your passwords..
You stay in control. 1Password for Agents free and optional for personal and family accounts. Unless you opt-in to an agent connector, agents cannot touch your vault.
ā”ļøLearn more: bit.ly/487da6J
47
I got hacked on multiple fronts this week. one of them appears to be Chinese but unclear. sht was and is still stressful having to go through all projects, patch them up, harden, and run security tests with as many edge cases, while also creating watchdog agents for monitoring. my roomateās school also got hacked with ssn numbers and other PII data breached. so once i started seeing the php and node vulnerabilities, it all started making sense. a new era for cybersecurity. buckle up.























