DearVCs, CEOs, VPs & Governments: Many people, including some AI researchers, thought this video generation and other AI advances would be impossible. Some researchers and engineers persevered with optimism, thought hard, derived novel mathematical algorithms and tractable

Somebody should have told Inflection AI that's an option 😉☺️☺️

Very nice to see Qwen ditching the mess that MMLU is, and using our reannotated MMLU-Redux for evaluation 🔥 arxiv.org/abs/2406.04127 huggingface.co/datasets/edinb…

I have joined the LLM pre-training team at poolside ⛱️ where we are building the world's most capable AI for software development. It's very exiciting to see how far the team have come already, and the recent 500M Series B will only accelerate this further.🚀

.poolside will be at re:Invent next week – if you’re in Vegas and want to see what we’ve been doing in the last 18 months, I’ll be on stage with Anthony Liguori at 2:30pm local time next Tuesday. More details here → poolside.ai/reinvent ⛱️

is deep learning hitting a wall? we have good news in time for thanksgiving :-) after another round of scaling law speculation, Eiso Kant (co-founder and cto of poolside) and i recorded a 45 min discussion for Air Street Press. we talk about scaling laws, synthetic data,

Announcing poolside's partnership with Amazon Web Services 🔥

Very cool to see DeepSeek also adopting MMLU-Redux, after Qwen used it for evaluating Qwen too. Nicely done Aryo Pradipta Gema👏 📃 arxiv.org/abs/2406.04127

MMLU-Redux will be at #NAACL2025, go talk to Aryo Pradipta Gema if you are there. 👏

Many people commenting on DeepSeek are hugely underestimating how much more compute it will take to get to AGI

Super excited to share 🧠MLGym 🦾 – the first Gym environment for AI Research Agents 🤖🔬 We introduce MLGym and MLGym-Bench, a new framework and benchmark for evaluating and developing LLM agents on AI research tasks. The key contributions of our work are: 🕹️ Enables the

According to Eiso Kant from poolside - human-level AI in knowledge work could be achieved within 18-36 months. His vision is to dramatically increase software development productivity and accessibility. On MLST now.

I'll be at ICLR 2026 in Singapore 🇸🇬 for poolside - not presenting any work, but looking forward to many interesting talks, discussions and catching up with everyone!

MMLU-Redux just touched down at #NAACL2025! 🎉 Wish I could be there for our "Are We Done with MMLU?" poster today (9:00-10:30am in Hall 3, Poster Session 7), but visa drama said nope 😅 If anyone's swinging by, give our research some love! Hit me up if you check it out! 👋

We’ve released our research vision. The key insight (which has been the through line for poolside since the beginning): Reinforcement Learning is the most important scaling axis to AGI In a world full of noise about AI's future, we have a specific view on how we’ll get to AGI

So many people I speak to don't find their job/the work they do interesting. Made me realise once again the privilege I have to be excited about the work I do, so many cool experiments to run and things to implement every day poolside

Stop overfitting to GSM8K! Reasoning Gym - 100+ RL envs for LLM RL - got accepted to NeurIPS as Spotlight! Frontier LLMs still struggle with many hard env configs. arxiv.org/abs/2505.24760 github.com/open-thought/r…

We believe that to compete at the frontier, you have to own the full stack: from dirt to intelligence. Today we’re announcing two major unlocks for our mission to AGI: 1. We're partnering with CoreWeave and have 40,000+ NVIDIA GB300s secured. First capacity comes online