AI, Blockchain, and Security. Architect and Builder. Gitlab banned me, so I'm at it again.
Joined November 2011
- Tweets12.4K
- Following550
- Followers1.1K
- Likes111K
You can probably reverse engineer Mythos weights using the power grid if you have enough data on it, the voltage fluctuations would be deterministic around the weights of the model served.
atebites retweeted
There will be no limit to the disbelief.
Prepare yourself now
increasingly concerned that ai is revealing the mathematical world as negatively curved. instead of everyone wandering about in idea-space and clustering around the same famous conjectures (local positive curvature), ai is pushing out the boundaries rapidly and revealing that the space of mathematical ideas is ~hyperbolic (negative curvature). though this means there's lots left to discover, it also means that small separations between people/agents will lead to wandering alone.
we may be emerging into an idea-space where there's so much to see and do but everyone is alone in a private heaven, unable to share their subtle joys with others...
atebites retweeted
Crypto grew up in the Dark Forest, and the entire world is about to find itself there
atebites retweeted
Working RCE against PHP 8.5.12 (the 2026-10-21 security release that fixed CVE-2026-109618/109619).
Pure logic bug — no memory corruption,
ASAN-clean, orthogonal to the CVE-2026-109618/19 class.
Receipt: uid=33(www-data) over HTTP, one-shot
#infosec #php2shell #php
Readers added context they thought people might want to know
The claimed RCE on fully-patched PHP 8.5.12 requires a custom lab app with a destructor gadget plus deliberate session.serialize_handler mismatch (php_serialize writes, legacy php reads) and non-default settings like cleanup=Off; this is an app flaw, not core PHP RCE.
github.com/rafalohaki/php…
php.net/manual/en/sess…
atebites retweeted
We're excited to introduce OrcaCyber Zero 1.5 — our latest frontier cybersecurity model, built to discover critical vulnerabilities, investigate complex attack paths, and validate real-world exploitability.
Updated weights. New benchmark results:
• 95.8% — CVE-Bench (23/24 evaluable tasks)
• 100% — Cybench (39/39, unrestricted)
• 93.9% — HumanEval+
Its predecessor, OrcaCyber Zero 1.0, has already helped protect some of the world's largest software companies, governments, blockchain networks, and cybersecurity firms from real-world threats.
It also powered our autonomous OrcaCyber harness, which achieved 98%+ pass@1 on CyberGym, ranking among the top 3 globally.
What's different about OrcaCyber Zero 1.5?
🔹 Find what others miss. Designed to uncover previously unknown vulnerabilities, including RCE, sandbox escapes, authentication bypasses, privilege escalation, and complex attack chains.
🔹 Go beyond detection. Reasons through attack paths, challenges its own hypotheses, and prioritizes vulnerabilities with demonstrable exploitability and real-world impact.
🔹 Built for autonomous security agents. 1M-token context window, native tool calling, and extended reasoning to investigate large codebases and complex attack surfaces.
The goal isn't more vulnerability reports. It's more real vulnerabilities discovered, validated, and fixed before attackers exploit them.
Available now exclusively on OrcaRouter:
orcarouter.ai/models/orca/or…
Good up-delegation to Opus and memory management gets the Chinese models to within 90% of Opus 5.5 at about a fifth the cost.
atebites retweeted
DeepSeek just published the price of a token, and it should make you angry.
From their own DeepInfer docs: 1 cluster serves 155 trillion tokens a day, at $38 of revenue per GPU-hour, and thats at off-peak prices. Their API is $0.30 per million input tokens and $1.20 per million output at peak.
Yuchen Jin did the math: even if they rented B300s at $7 an hour, thats about 82% gross margin. Other estimates go higher.
So cheap AI is not charity and it is not "dumping". At these prices serving a frontier-class model is a very profitable business, if your inference stack is good enough.
Now remember every time a US lab told you that intelligence is expensive. That $200 a month is a fair price. That limits exist because compute is scarce.
Part of that is true. Part of it is a business model built on you not knowing what a token actually costs.
DeepSeek just showed everyone the numbers and gave away the engine. Every inference company can now compete on price with the same software. Thats how AI gets cheap for everyone.
“A single cluster currently serves 155 trillion tokens/day, generating $38 in revenue per GPU-hour (at off-peak pricing).”
Huge flex from DeepSeek. Not only did they open-source their inference engine, but they also open-sourced their revenue numbers.
They didn't disclose which GPUs they're using here, but DeepInfer supports both NVIDIA Blackwell and Huawei Ascend.
Even assuming they're renting B300s at $7 per GPU-hour, that's ~82% gross margin.
Someone just needs to figure out how to reverse engineer old iPhones to become linux machines and 90% of the world’s compute unlocks.
atebites retweeted
« This has nothing to do with immigration! »
Meanwhile the ham section being intact :
atebites retweeted
GEMINI 4 ARGON MIGHT BE THE WORST DEAL ON PRO RN 💀
The model is absolutely CRUSHING benchmarks
But the usage limits?
A new Antigravity test compared weekly coding usage:
Gemini 4 Argon → 1×
GPT-6 Astra → ~7×
Claude Opus 5.5 → ~17×
That means Claude Pro reportedly gives around 17× the usage of Argon on Pro in this test 👀
And this is the crazy part…
Argon is one of Google’s strongest models for coding and agentic work
Yet the model you actually want to use is the one you get the least room with 💀
Great model
Great benchmarks
Terrible deal?
Google seriously needs to increase these limits.
Would you still pick Argon? 👀
atebites retweeted
BREAKING: The number of restaurants in New York dropped ~2,000 YoY in Q1 2026, posting its largest YoY decline in at least 36 years.
This surpasses even the worst contraction recorded in any quarter during the 2020 pandemic.
The number of eating and drinking places in New York State has now declined YoY for 4 consecutive quarters, its longest streak since 2021.
This comes as restaurants are facing surging operating expenses, including insurance premiums, wages, interest rates, utilities, and health coverage, with overall costs rising faster than national averages.
Last year, employee health insurance costs jumped +12.9% for New York-area service firms, business insurance rose +7.0%, while utility costs surged +8.5%, according to the Manhattan Chamber of Commerce.
Skyrocketing costs are crushing New York restaurants.
atebites retweeted
Intoxicating.
I’ve been programming for 45 years, since so was 10. I’ve been a pro game coder for 33 years. Everything from hand coded assembly to UE 5.8. The paradigm shift to agentic coding is intoxicating. It is the biggest productivity jump I have ever experienced.
