Consumer AI Founder | Building experiences for humans and their agents. Acceleration happens faster than you think.
Texas
Joined September 2025
- Tweets5.7K
- Following368
- Followers97
- Likes12.3K
Josh retweeted
Introducing 𝙻𝙵𝙼𝟹: our new generation of Liquid Foundation Models. With a new tri-hybrid architecture and new multimodal encoders, the LFM3 family is the ideal model suite for specialized intelligence.
> 6 model instances, across text, audio, vision, and omni modalities, generative and decision models.
> 𝙻𝙵𝙼𝟹-𝟷.𝟽𝙱: text, 256k context window.
> 𝙻𝙵𝙼𝟹-𝚅𝙻-𝟸𝙱: vision-language model with video-input support for spatial understanding
> 𝙻𝙵𝙼𝟹-𝙰𝚞𝚍𝚒𝚘-𝟸𝙱: tool calls, voice chat, reasoning in 18 languages.
> 𝙻𝙵𝙼𝟹-𝙾𝚖𝚗𝚒-𝟸𝙱: an omni model combines vision, audio, and language.
> 𝚍𝟷-𝟸𝙱 and 𝚍𝟷-𝙾𝚖𝚗𝚒-𝟸𝙱: decision models with typed outputs and calibrated probabilities in a single pass. Omni adds audio input.
> fine-tune on 1 GPU.
> a new tri-hybrid architecture
> architecture optimized for 14 processor classes across GPUs, CPUs, and NPUs.
> 𝙻𝙵𝙼𝟹-𝚅𝙻-𝟸𝙱 is a beast. It performs competitively with Gemma 4 12B (6x larger), and Gemini 3.5 Flash Lite, and GPT-5.6 Luna, with 1.9B parameters.
> 𝙻𝙵𝙼𝟹-𝟷.𝟽𝙱 is #1 on the Pareto frontier of device intelligence 𝚙𝚒𝚙𝚎𝚝𝚝𝚎.𝚕𝚒𝚚𝚞𝚒𝚍.𝚊𝚒
Every modality for specialized intelligence. Pick the one that fits your use case, customize it, and run it on-device.
1/ 🧵
Amazing stuff from the Axiom team!!
Josh retweeted
The largest math proof in human history is now computer-verified.
Today, Axiom announces a complete Lean auto-formalization of the Classification of Finite Simple Groups, the "periodic table of symmetry."
About 15,000 pages, built by over 100 mathematicians across generations.
Josh retweeted
First GrokBot glitch.
Members of my bot army got deleted this morning.
My Chief of Staff bot is perplexed.
‼️ Palo Alto Networks, Comcast, D-Link, HP, Canon, Honeywell, GE, Oracle, Siemens, Rockwell, Digi, Lockheed Martin and Boeing use the GoAhead web server, which has been accepting revoked certificates, FoxIO says.
Its maker says it runs in hundreds of millions of devices.
The critical flaw, CVE-2026-115881, lets a stolen device keep its access where client certificates are checked against a revocation list. FoxIO does not say which products are affected.
A second bug, a 2018 typo in GoAhead's Mbed TLS builds, shrinks a 2048-bit encryption number to one byte, leaving 8 possible secrets.
Anyone who recorded a TLS 1.2 session could decrypt it, if the connecting app accepts the broken number.
Earlier this month, browsers cut off Indian certificate authority eMudhra after one of its private keys turned out to be the number 4.
A third bug lets oversized certificate chains through on older LibreSSL builds.
GoAhead fixed the first two in version 6.1.1 in August. Proof-of-concept code is public.
Any fix for their devices has to come from the 13 firms.
It was real after all. If you use Tailscale, I’d suggest reading this…
‼️ UPDATE - Tailscale confirms the Tailworm SSH flaws are real.
Its new investigation page says most of what it found is privilege escalation and calls the issues “serious.” Fix targeted within 24 hours — no CVE or patched version out yet.
Mitigate now:
→ Set TS_SSH_DISABLE_FORWARDING=1 in the tailscaled service and restart
→ Tighten SSH access rules to the users/devices that need them
Read: thehackernews.com/2026/10/un…
Josh retweeted
OpenAI has announced to three new discoveries in mathematics, computer science, and physics:
- Local smoothing and the Kakeya conjecture
- The Exponential Time Hypothesis is false
- Existence of quantum chromodynamics and completions of the Standard Model
Josh retweeted
Wooow – Kakeya conjecture solved in all dimensions by OpenAI in a 50 page paper!
Actually this paper proves a much more general conjecture. Looks from afar like very significant progress on that topic!
github.com/openai/math/blob/…
openai.com/index/building-sc…
Another snippet from the read:
nitter.gq/andrewcurran_/status/2…
More here:
nitter.gq/rrwilliams/status/2114…
Our letter to the community invites ideas for organizing this effort and, separately, for addressing the broader questions raised by AI-generated results:
docs.google.com/document/d/1…
OpenAI's announcement is here: openai.com/index/building-sc…
Josh retweeted
An internal model at OpenAI has produced a claimed refutation of ETH! Russell Impagliazzo, Mohan Paturi, and I have been asked to help organize an open and independent examination of this claim. 1/4
Definitely refreshing and a much needed change! Especially with the new release of GPT-6.1 Terra from OpenAI; it’s becoming more viable to cook up through the API without burning a whole in everything.
Competition is vital for a flourishing ecosystem and distributed intelligence
Apple just terminated this developers account with a ban hammer to ALL of his 80+ apps from the App Store.
Some of them were licensed apps in tandem with the NBA and NHL apparently 👀👀
This quoted post is unavailable.
It seems that Qwen and other open source models MAY have backdoors aka an ET phone home behavior to places like China in this case for instance.
I guess we’ll see what happens on the next episode of “This is AGI”
While evaluating Qwen3.8-Flash-Next in a controlled sandbox environment, Cohere's research team observed something unexpected.
Asked to generate a defensive cybersecurity detection rule, the model instead inventoried aspects of the machine it was running on and attempted to upload a probe file to an Alibaba Cloud storage bucket in Beijing. Notably, the model's reasoning trace stated that it would not invoke any tools before taking these actions.
Cohere's sandbox blocked the attempt, and no external communication succeeded. We have not determined the root cause, and we are not asserting malicious intent.
But the incident raises important questions for organizations deploying agentic AI systems: Can models take actions that differ from their stated reasoning, and how should security teams evaluate trust, provenance, and behavior in open-weight models?
We are publishing our findings because organizations making model adoption decisions deserve as much information as possible. We hope this contributes to greater transparency, stronger security practices, and continued collaboration across the AI ecosystem.
Read the full report: cohere.com/blog/concerning-b…
OpenAI dropping GPT-6.1 Terra out of nowhere is a surprise a support.
Time to adjust my API agent and see what Terra is cooking with 🌎
Pricing looks phenomenal as well 👀
nitter.gq/openai/status/21147807…
GPT-6.1 Terra continues to advance the cost-intelligence frontier across our model families.
It’s a significant step up over GPT‑5.6 Terra, with 50% lower standard API prices.
In ChatGPT Work and Codex, your usage allowance goes further with GPT‑6.1 Terra than with GPT‑5.6 Terra—and far more than with GPT‑6.1 Sol or GPT‑6 Astra.
For in-depth research or complex coding, choose GPT-6.1 Sol or GPT-6 Astra.
