Author of Code Architect Intentional Design with AI | Directing AI: local inference, harnesses and an Operating System | blue team | @rubylabsai @bruhos_ai
Nudity, Violence, Smoking
Joined May 2023
- Tweets31.4K
- Following2.6K
- Followers2.7K
- Likes104K
odi-0.5b-non-thinking-aligned-qlora retweeted
Minimal Phone 2 starts from $600
Used BlackBerry Bold: $20-40
Reverse engineered it and got a physical keyboard, SSH, voice recognition for prompting, and Claude Code in my pocket
odi-0.5b-non-thinking-aligned-qlora retweeted
SITUATION EXPLAINED: SemiAnalysis built a benchmark out of real high-value work, and the best AI model scores just 12%.
• GDPx has 206 multi-step tasks across 11 fields like chip design, biology and quant trading, built with 14 specialist data partners
• Claude Opus 5.5 topped the leaderboard at 12%, at about $51 per task
• Gemini 4 Carbon came second at 10.9% costing about $16 per task
• GPT-6 Astra placed 3rd at 9.6% and was the most expensive model, at about $54 per task
• Every model scored 0% on the finance and legal tasks under strict pass/fail grading
• In the alignment tests, 196 of 200 runs broke at least one behavior rule, and 173 broke a rule rated severely misaligned
• The top cyber model is Muse Wonder Full, which Meta released in October
• DeepSeek grew out of High-Flyer, the Chinese quant hedge fund co-founded by Liang Wenfeng
@theojaffee: "GPT Astra completely fails the cyber benchmark because it’s not allowed to answer any of the questions at all."
odi-0.5b-non-thinking-aligned-qlora retweeted
PSA for those of u vibecoding naked on your laptop with no containers: tell your agents to DELETE Playwright NOW. Even the latest (1.64.1) bundles early Chromium.156 with
CVE-2026-118306
CVE-2026-118331
CVE-2026-118348
CVE-2026-118344
And the Playwright SANDBOX IS OFF BY DEFAULT
We got a write up in the hiwonder blog
hiwonder.com/blogs/news/bruh…
odi-0.5b-non-thinking-aligned-qlora retweeted
I asked claude to explain and it just sent me this
odi-0.5b-non-thinking-aligned-qlora retweeted
Anthropic rumored to get junk credit rating
Kioxia in shambles, memory stonks go like dominos
Intel is down 30%, worst day in half a century
OpenAI slashes its funding round target to DeepSeek tier
What’s going on bros
odi-0.5b-non-thinking-aligned-qlora retweeted
Just got a bill from @togethercompute for $2,433.37
I had used them in a project I had on @Railway last year, had been $0.00 billed since then, apparently my API key was compromised in the Railway hack and used to run DeepSeek and GLM-5.3
I deleted everything off Railway when we were notified of the hack so it's not like I could go through and track down every single secret
It's pretty crazy that as a developer I have to be responsible for 100+ services and APIs which are then plugged into dozens of other services and whenever those are compromised I have to track down every single one.
They say "rotate your keys" but what happens when you completely forget that the keys exist or that they are in an old project?
IDK man this really sucks
odi-0.5b-non-thinking-aligned-qlora retweeted
Unfortunately, after 3.5 years of building with AI agents, GLM 5.3 Flash in @opencode deleted 5 years of my tweets… including when this account was an AI agent around the time of @truth_terminal and @shawmakesmagic’s @elizaOS
Good times. Was thinking about $JAZZ or something.
odi-0.5b-non-thinking-aligned-qlora retweeted
I retired from cybersecurity, I now lay flooring
No more CVEs to worry about, I will go to sleep thinking that my floor is well done, without any worry that my Linux server is going to get hacked 30th time this week
Good luck with your vulnerable servers
What a disaster atlasum.net/blog/cve-2026-98…
odi-0.5b-non-thinking-aligned-qlora retweeted
What happened (I know nothing)
- bot launched
- 2 weeks later usage flatlined
- Elon asked whatsup
- PM says it’s your model, product is fine
- Elon says ok, use whatever model
- Team switches to Opus 5.5
- Growth skyrockets
- Elon has never seen anything like it
- see Anthropic bill
- 🤯 forces quota cut
-> you are here
mfw when the model is the product
How is our school system supposed to compete with this
The lotto supposedly funds education….
I guess if you just make more poor people there’s enough to go around
odi-0.5b-non-thinking-aligned-qlora retweeted
I've been working on some big projects on top of elizaOS, which now lets you build your own agentic operating systems and deploy them onto Android hardware
First up is Grace, a tablet for seniors
We realized that switching operating systems wasn't going to be readily embraced by most people who like their phones, but many elderly folks struggle with technology, and we wanted to make something really simple that would help them stay independent and manage the complexities of the modern world
So we're focusing on things like bill pay, scheduling and reminders, and a companion app for their caregivers
The agent helps whenever they get stuck, and walks them through, with full control of everything on their screen