


building agent tools 🤖 oh-my-pi contributor · tinkering centaur.run · prev @uber. money legos are still cool. 👩🚀 @egirl_capital · 👨🌾 @199xcapital


Some of you have been asking very politely. Some of you have just been asking. GPT-6.2 Astra and Sol are now available. Super efficient and built to perform where it counts—the real world. openai.com/index/introducing…

introducing ghostmotion g1 10 mins of video in 1 prompt

Someone verified wintermutes cexdex contract on etherscan lol. Onchain is now open source etherscan.io/address/0x51c72…

cc @_can1357 @BwitneyHouston @wolfie_ @banteg @usr_bin_roygbiv @itomatik h/t @Dazzuh for the help investigating and kind words 🙏

the tradeoff: all-Drex made find ~4× slower (p50 0.69 → 2.97s) find.verifyModel lets Jev handle fast routing while Drex verifies passages find.hitThreshold lets you tune the cutoff per model ✅ github.com/can1357/oh-my-pi/…

i replayed 1,100 real `find` verification requests (100 queries, 711 files). two blind graders agreed on 636 files (κ=0.79). at the same τ=0.20 cutoff: jev: 51% precision / 99% recall drex: 75% / 97% AUC: jev 0.913, drex 0.965, @liquidai d1-ultra 0.980.

omp's find is a staged context-selection pipeline: judges rank filenames/sketches, then verify full passages answer the query. scores shape what coding agents see 👀 across 14d (~1K searches, ~6K hits), agents read 65% of hits scored 0.90–1.00 vs 8% at 0.20–0.35.

shipped some find improvements in omp v18.20.0 🚀 you can now use separate judges for fast filename/sketch routing and stronger passage verification, with native support for @NaceAI's Drex 🔥 why does this matter? the best coding model can't fix bad context. 🧵

telegram has shipped multiple rces already in the new widget engine, with code execution demonstrated on android, ios and windows. here’s how to outrun the inevitable macos rce: gist.wavey.info/AB9t7eBoQhyx…

I have a major result to share. I’ve found conclusive forensic evidence that the top 4 Chinese AI labs—DeepSeek, Moonshot, MiniMax and Ant Group—have built their newest models on NVIDIA’s most advanced Blackwell chips, which are barred from export to China. This sounds insane, but I discovered that almost every AI model carries a hidden signature of the GPU that trained it. When a model is first set up, the starting values of its weights are laid out in a pattern that depends on the exact GPU. Each combination of GPU and seed number produces a unique pattern—a sort of fingerprint of the hardware used to create the model. Traces of that pattern survive training into the published model. Again, this evidence isn’t based on any insider information or leaking. It’s based on a forensic analysis of the model weights released by the companies themselves. I searched more than six trillion combinations of chip layout and seed number and matched them to open-weights models. In doing so, I found conclusive evidence that six Chinese AI models—including DeepSeek-V4-Pro, Kimi K3, and MiniMax-M3—were initialized on NVIDIA Blackwell systems. Four of the models match Grace Blackwell, the GB200 and GB300 parts in NVIDIA’s flagship liquid-cooled rack-scale systems. I then verified these results on real hardware. Running unmodified PyTorch with the seeds I’d recovered, a Blackwell B300 reproduced DeepSeek-V4-Flash’s untouched starting weights bit for bit, and a rented GB200 reproduced Kimi K3’s to a correlation of 0.999995. A Hopper H100 reproduced none of DeepSeek-V4-Flash’s, and changing the seed by one broke the match. (DeepSeek’s seed is 953049583 by the way.) The method also was verified to match 24 known open-weights models to the hardware that developers said they used. It is exceedingly unlikely that these results are a coincidence. Across six trillion combinations, the best accidental match should reach only about 8 sigma (i.e., eight standard deviations above what random chance would produce). The real matches range from 49 to 1,385 sigma, and even the weakest single test hits 20. These were data-center-scale training runs. With some analysis of release schedules and model details, I estimate that the largest, Kimi K3 and DeepSeek-V4-Pro, used 2,000 to 3,000 Blackwell GPUs each. Clusters of this size cost on the order of $150 to $200 million each to buy, and across these labs alone would add up to more than half a billion dollars of export-controlled GPUs, not counting experimental compute or otherwise. It’s quite likely that we see multi-billion dollar expenditure on export-controlled GPUs. I am highly confident in this result, and it has been independently verified by senior researchers at OpenAI and Anthropic. You can read the full report at unemploymentlabs.com/checkpo…. There is much more I’ve identified about these models through other forms of checkpoint forensics like this, and will have more to share soon.

sokobanteg v0.2 live gpt-6.1 sol, gpt-6.1 terra, gpt-6 astra claude opus 5.5, claude sonnet 5.5, claude haiku 5.5, claude fable 5.2 gemini 4 argon, gemini 4 carbon qwen3.8 max 0902, kimi k3.1 pro, muse wonder, muse ember, grok 4.7 build x.com/i/broadcasts/2ZBjVkaLm…

1/ IRIS goes live in a few hours, and we’re opening in phases. The first is Core Phase - a short list of collateral, two stablecoins and set limits, which we’ll widen as we go. Here’s what’s in it :

new fav models just dropped 🔥
Introducing Claude Fable 5.2 and Claude Mythos 5.2, designed to tackle the most daunting technical challenges. They’re capable collaborators on ambitious projects in many fields, and show improvements in areas like coding, reasoning, and vision.

INFI Airdrop Registration is NOW LIVE. The Season 1 & 2 snapshots are complete. Your next step toward the INFI airdrop starts here. Register your wallet, select your terms, and prepare for TGE. Registration Closes: Nov 10th at 15:30 UTC Register now: app.infinifi.xyz/infi

SOLAR: 100x FASTER SOLIDITY Solar is the future of Solidity and the EVM. It's >100x faster than Solc. It has faster bytecode than Solady with no assembly. Its bytecode is faster to interpret by the EVM. Solar fixes Stack Too Deep! foundryup forge --use solar Blog below.