🚀 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
A trader who’s compounded at 60% a year with a Sortino over 2 on why he refuses to compete with the pods on their timeframe:
Joe Walsh is the Founding Partner of hedge fund Code Partners.
Joe Walsh explains:
“The first thing I think about is time horizon.”
“I don’t want to be valuing a security on a three to six month timeframe where I’m competing against a very well-resourced, motivated person inside one of these firms with way better corporate access, way better short-term information, proprietary data sets.”
“You’ve got to pick your time horizons to be avoiding very good players. You’ve got to pick your sectors where all the pods who used to do that have been shut down and all the investors have given up.”
“The game selection aspect is running away from competition, and that’s where greater inefficiency by definition lives. There are huge pockets of that in the equity market today.”
“If you think you’re going to have a huge edge trading European industrials on a three-month timeframe, I would politely disagree.”
“We have huge edge in timeframes greater than 12 months, in sectors and markets that are just not as competitive. It’s not worth their time, when they’re running $100 billion levered multiple times, to muck around in names with $10, 20, 30 million a day of liquidity.”
“The capacity I could run money with is very attractive, to single digit billions, and still have a huge edge and still remain very liquid.”
🔍 Search & Fetch APIs are now open to everyone!
Powered by our in-house search engine, serving hundreds of millions of requests daily worldwide.
💰 Each endpoint: $0.75 per 1,000 calls.
📖 Docs: https://t.co/xLVnK4QQIr
📌 Search & Fetch calls via server-side tools in DeepSeek Harness, Claude Code, and Codex are currently free for DeepSeek API users. Billing starts Oct 31, 2026, at 16:00 UTC.
3/4
FINALLY found how my key got leaked and got the $90,000 bill from google
Opus 5.5 said codex had PUBLISHED MY ****ING .ENV TO CLOUDFALRE PAGES ‼️
Meanwhile codex couldn't find anything, refused to explain itself and flagged my chat for cybersecurity risk lol
‼️ The Arch Linux drama continues: an account uploaded at least 15 fake copies of yay, Arch Linux's popular package helper, with malware in their install scripts.
A scan of one, yyay, shows it pulling code from a Russian git host and running it as root.
i am short companies whose margins depend on the cost of performing verifiable work.
i am long companies whose margins depend on owning very very consequential outcomes.
this applies to both public/private markets.
🚨 CHINA JUST OPEN SOURCED AN OCR THAT EATS 100 PAGE PDFS IN ONE PASS
It’s called Unlimited OCR. Only 3B params. Runs locally.
Every other OCR tool chops your doc into pages and loses the thread. This one reads the whole thing in one continuous pass.
→ One shot full context parsing (32K context window)
→ Multilingual out of the box
→ 93% on the standard parsing benchmark (6 points over baseline)
→ Under 0.11 error rate past 40 pages
→ Runs 100% locally on your own hardware
→ Works with Transformers, vLLM, SGLang, Docker, Ollama, llama.cpp
Traditional cloud OCR (Textract, Google Vision, Azure Doc Intelligence) costs $1.50 to $15 per 1,000 pages.
This runs on your machine. For free. Forever.
Baidu built it explicitly to push DeepSeek OCR one step further. Already at 1.9M downloads on Hugging Face and most people have no idea it exists yet.
100% open source.
Reverse engineered my AC, and vibe coded a lil status bar app so I don't have to find the remote every time.
No first world problem will remain unsolved in 2026.