E_genius

building 🐶 app for dog families
+ 🦉 AI contract review for lawyers

[E_genius @yevloop]

ok X, real talk 🙏

i lost my dog.
so i started building 𝗖𝗵𝗼𝘄𝘆 🐶 so other dog families don't miss the signs
https://t.co/xFksURk8Ma

solo, from 🇺🇦, on Gemini Pro
next: wearables, an oura for dogs 🐾

i wish i could build it with 𝗔𝗿𝗴𝗼𝗻 on Ultra
and post every step here, for free, for everyone

@OfficialLoganK @GoogleAI if this reaches you 👀

one RT might get it there 🔁

[E_genius @yevloop]

@ExactlyJosh @XFreeze thank you!

[E_genius @yevloop]

This is why I love the 𝕏 community.

I wasn't expecting anything. Just sharing my journey.

And today, thanks to @XFreeze , I got a free month of Super worth 120$!

Sometimes, a small act of generosity can mean the world to someone trying to build something from scratch.

This is the kind of community that makes you want to keep building, sharing, and giving back.

Thank you so much, @XFreeze .
You genuinely made my day!

[E_genius @yevloop]

Mostly to guide AI and verify its output. It's already heading in that direction.

I've almost completely stopped writing code by hand. Now I mostly review AI-generated code, verify that it works correctly, and guide the AI on what to build and how it should work.

I feel like we're slowly shifting from writing code to managing AI that writes it for us.

[E_genius @yevloop]

openai gpt-rosalind just found a new drug discovery lead

thousands of agents read gene data from millions of mouse brain cells in 32 minutes
and pointed at cholesterol transport in neurons as a mechanism to test

now it needs lab validation. curious how many of these leads hold up

[OpenAI @OpenAI]

GPT-Rosalind surfaced a new lead for drug discovery: could restoring cholesterol transport protect neurons from degeneration?

DCTN1 is a gene linked to Perry syndrome, a rare neurodegenerative disorder. In mouse-brain data, disrupting it increased expression of genes involved in making cholesterol. GPT-Rosalind suggested a possible explanation: cholesterol may not be reaching where neurons need it.

That gives scientists a specific mechanism to test and, if confirmed, a direction for developing new treatments for neurodegenerative disorders.

With @perturbai_tx and @scrippsresearch, a team of thousands of GPT-Rosalind agents worked in parallel, connecting gene-expression results from millions of mouse brain cells with scientific literature in just 32 minutes. We’re now expanding this collaboration to identify new drug targets.

https://t.co/oHdF3ELsMj

[E_genius @yevloop]

@omarsar0 self-evolving harnesses sounds promising. does the gain hold up when you swap in a weaker base model?

[E_genius @yevloop]

@xlr8harder the function naming thing is real, i still have to look up argument order half the time. vibe coding at least stays consistent

[E_genius @yevloop]

@ObbeVermeij wow, a crashing mess for most of development and it still became that. what finally turned it around?

[E_genius @yevloop]

@synthwavedd honestly agree, image gen has felt stuck behind that one for a while. curious what the bigger google model ends up doing

[E_genius @yevloop]

@thtbee_ the small bee it added on its own is such a nice touch. did you give it any reference or just the prompt?

[E_genius @yevloop]

@igorlachenkov @StephenKing @StephenKing

[E_genius @yevloop]

openai patch the planet: ai found critical security flaws in PHP

php runs an estimated 70% of websites
fix took two months with a coalition of maintainers and needed a major change to the php engine
openai says frontier ai found it ahead of attackers

if you run php, upgrade. curious how many sites actually will

[OpenAI @OpenAI]

OpenAI models helped researchers—operating as part of our Patch the Planet initiative—discover critical security flaws in PHP, an open-source programming language used to run an estimated 70% of websites.

Over the past two months, we led a coalition of maintainers, researchers, and security providers to prepare fixes and protections ahead of disclosure. This duration was necessary because the fix was quite intricate and required a major change to the PHP engine. Every site using PHP can now safely upgrade their version to stay secure, in addition to applying mitigations from various security vendors.

Due to the severity of these flaws and PHP’s widespread use, we consider this bug to be one of the most impactful vulnerabilities in recent memory. It also marks a new phase, where AI can find security bugs with internet-scale impact.

This is an example of the Defender’s Window in action: to our knowledge, frontier AI was able to find this vulnerability ahead of attackers, giving defenders time to prepare.
https://t.co/TW3jrSbB3f

[E_genius @yevloop]

mistral voxtral 3 is out
state-of-the-art transcribe and speech models
powers mistral's new voice agents api
curious if open weights come with it, voice is getting crowded fast

[Mistral AI @MistralAI]

Introducing Voxtral 3: state-of-the-art Transcribe and Speech models powering Mistral’s new voice agents API. 🧵

[E_genius @yevloop]

deepseek just open sourced DeepInfer, the inference stack they run in production

same thing powering V4.1-Flash at scale
high throughput, low latency, lower compute costs

as a solo builder i love this part
labs sharing the boring infra, not just the weights

who's self hosting this first?

[DeepSeek @deepseek_ai]

🚀 Introducing DeepInfer, our in-house inference framework, now fully open source!

Powering DeepSeek-V4.1-Flash at scale with high throughput and low latency. We’re sharing our production stack to help the community lower compute costs, reduce energy use, and make AI more affordable.

🔗 https://t.co/JGAik1kgk0

1/4

[E_genius @yevloop]

mercury 3 just dropped. diffusion llm, 1,000+ tokens per second

inception says it beats gpt-6 luna and claude haiku 5.5 on a range of benchmarks
40x bigger than mercury 1, 1M context, already on openrouter

as a solo builder the speed part is what gets me
instant replies change how an app feels

anyone tried diffusion models for real agent loops yet?

[Inception @_inception_ai]

Mercury 3 is live: the most capable diffusion LLM released to date.

It outperforms GPT-6 Luna and Claude Haiku 5.5 across a range of benchmarks while running at over 1,000 tokens per second.

More than 40x the size of Mercury 1, 3x more intelligent across standard benchmarks, and just as fast, with a 1M-token context window.

Available today on the Inception API, @baseten, and @OpenRouter .

[E_genius @yevloop]

minimax m3.1 flash — space bunny was them the whole time

200t+ tokens on the boards as a mystery model, now it has a name

frontier agent scores at flash prices. weights said to land oct 23

anyone actually shipping with this yet, or still all deepseek for cheap agents?

[MiniMax (official) @MiniMax_AI]

MiniMax M3.1 Flash: Space Performance + Bunny Speed

🚀🐰 Yes, Space Bunny - which has topped global leaderboards with 200T+ tokens - was an earlier version of M3.1 Flash.

💰Redefining the performance economics of agentic AI. M3.1 Flash achieves frontier-level intelligence across key tasks such as Cowork (Automation-Bench: 62%; ALE: 30%), Coding (DeepSWE: 74%) and Visual Reasoning (BabyVision: 91%), at 1/3 the cost of M3 and 3x the speed.

API pricing (flat rate): $0.15 / $0.60 per 1M input/output tokens, with cache reads at $0.015 / 1M tokens.

Model weights will arrive on Oct 23.

API: https://t.co/fHRdSV73Hr
M Plan: https://t.co/XfJWSyauda
MiniMax Code for desktop: https://t.co/SGmLTI2HQm

[E_genius @yevloop]

elorian glimpse is a vision model that goes back and re-checks the image while it thinks
zooms in, pulls video frames, measures with code
$0.60 per million input tokens
would you trust it to review your app screenshots?

[AI Scientist @AIScientist_X]

NEWS: ELORIAN RELEASES FIRST MODEL GLIMPSE AT $0.60 PER MILLION TOKENS

> Elorian AI released Glimpse on Oct 16 as a research preview through its API and chat.

> Pricing is $0.60 per million input tokens and $3 per million output tokens.

> Elorian says Glimpse beats GPT-6 Astra on four evaluations and matches it on embodied reasoning.

> The company says a small team built it with a few hundred GPUs on top of an open-weights base model.

> CEO Andrew Dai previously spent 12 years at Google DeepMind.

SOURCE: Elorian

[E_genius @yevloop]

openai aurora red, green and blue alpha just showed up in the api
no announcement, no pricing, people trying them get 503s
apparently the last aurora-alpha showed up right before a codex release
my guess is a fast coding model. what's yours?

[Andrew Curran @AndrewCurran_]

Three mysterious new models just showed up in the OpenAI API:

aurora-red-alpha
aurora-green-alpha
aurora-blue-alpha

[E_genius @yevloop]

@b4dgerr Congrats bro 🥳

[E_genius @yevloop]

@petergyang @bot @spacexai https://t.co/XsNUcthuY8

[E_genius @yevloop]

ok X, real talk 🙏

i lost my dog.
so i started building 𝗖𝗵𝗼𝘄𝘆 🐶 so other dog families don't miss the signs
https://t.co/xFksURk8Ma

solo, from 🇺🇦, on Gemini Pro
next: wearables, an oura for dogs 🐾

i wish i could build it with 𝗔𝗿𝗴𝗼𝗻 on Ultra
and post every step here, for free, for everyone

@OfficialLoganK @GoogleAI if this reaches you 👀

one RT might get it there 🔁