Fun week in NY for a cousin’s wedding

Generated some floral patterns last night #stablediffusion

Biggest waves I’ve ever seen here

The Devil’s Cat 🎃 💻 🎨

Nice to show my daughter a SpaceX launch on the beach her last night here

My neighbor got a drone and almost killed me haha

I typed some random things into the new text-to-video model, like it appears everyone else was doing today as well. I haven't seen much in this space for six months, but hopefully this first open model will accelerate things.

Summer fun in Georgia and South Carolina

Weekend project to use a local AI model to add color commentary to a simple networked tic tac toe game. Apparently the first move is "going for the jugular" lol. Model is just vanilla Llama 2 Chat 13b with a system prompt.

I wasn't expecting this chart to be all up and to the right, but the result was more drastic than I anticipated. According to Wikipedia, there are fewer 'notable' games per year in the early 2020s than the early 1980s? Cliff Bleszinski Tim Sweeney John Carmack

Giving a 5-yr-old an AI-enhanced painting program

First time I've seen a local model get the brothers/sisters question right. I only saw GPT-4 get it right very recently. Trying some early Qwen2 72B quants of Qwen 2 in moved to: @lmstudio

First announcement from our new startup, beating the new Gemini 3 model on ARC-AGI

Poetiq has officially shattered the ARC-AGI-2 SOTA 🚀 ARC Prize has officially verified our results: - 54% Accuracy – first to break the 50% barrier! - $30.57 / problem – less than half the cost of the previous best! We are now #1 on the leaderboard for ARC-AGI-2!

We have verified a new SOTA Gemini 3 Pro Refinement technique, authored by Poetiq 54% on ARC-AGI-2, $31/task

We finally had a moment to run our system with GPT-5.2 X-High on ARC-AGI-2! Using the same Poetiq harness as before, we saw results as high as 75% at under $8 / problem using GPT-5.2 X-High on the full PUBLIC-EVAL dataset. This beats the previous SOTA by ~15 percentage points.

We’re thrilled to announce a new chapter for Poetiq: We have closed $45.8M in Seed funding. It’s a privilege to build alongside partners who understand the scale of our vision, including Surface, FYRFLY, Y Combinator, 468, Operator Collective, NeuronVC, and HICO.

Following up on our SOTA results on ARC-AGI, we’re excited to share new SOTA results on Humanity’s Last Exam (both with and without tools) and SimpleQA! On HLE, Poetiq’s meta-system created multiple new SOTA configurations, going all the way up to 55%.