Introducing Prime Sandboxes: MicroVM sandboxes purpose-built for RL training. Model training requires running tens of thousands of concurrent sandboxes, leading to complex and costly configuration. We built Prime Sandboxes for our own team. Today we're releasing them publicly.

Introducing Pod 6. The sixth generation of our award-winning sleep system. ✅ The new Hub is small enough to fit under your bed frame ✅ More powerful than ever ✅ With 9x more sensors for enhanced accuracy ✅ In new Solo sizes, at our lowest starting price yet Get yours today at

which of your favorite token brokers are contributing back to open source ai? shall we count the ways? a slice of data: number of PRs upstreamed and merged into the most popular oss inference engines, by PR contributor org

We built high-throughput materials labs in Menlo Park to create a loop between experiments and models. The labs generate fresh data, the models learn from it, and then help us decide what to try next. Using only 1,300 H200s, plus months of our experimental data, we mid-trained and RL’d an open-source model to surpass GPT-6 Astra on our analysis benchmark. We call it Neon. This is real footage from our lab. We’re focusing first on hard problems in materials science, including superconductors, magnets, and semiconductor materials. Read our blog posts below.

Mercor now spends 3X as much on LLM inference as we spend on employee salaries. Our inference spend creates so much ROI that it’s additive to headcount, not replacing it. It’s becoming increasingly clear how we get to ~10% GDP growth: 1. Within 5 years, as AI diffuses throughout the economy, companies will spend as much on inference as they spend compensating knowledge workers today. 2. Knowledge-worker compensation is roughly $40T/year, so this would eventually mean ~$40T/year of inference spend. 3. If wages remain roughly constant and companies profitably absorb that much inference the way AI-native companies are today, the economy needs on the order of $40T/year of additional final economic output to support it. 4. Producing ~$40T more annual GDP in year 5 than a ~3% growth baseline implies ~9% annual GDP growth on average over the next five years. Based on the ROI we’re already seeing from inference at Mercor, this feels reasonable. AI-native companies are a leading indicator for the global economy.

my hot take is specialized inference engines are amateur projects to spend the spare tokens when @thsottiaux gives people one more reset😄 an inference engine is way more than just running a model on a certain hardware, it is an inference ecosystem. I use this slide several times when I talk to people about vLLM. Models, hardwares, and inference techniques, all of them move quickly. And vLLM lies in the intersection to provide a unified interface to end-users and applications. vLLM is the inference ecosystem. vLLM not only support current models and hardwares, but we are also working to support new models and hardwares coming in the next few months. It's an ecosystem people can trust and rely on. Over the past 3 years, I have seen so many projects claiming to be better than vLLM in certain aspects, but in the end either their techniques are contributed to vLLM or they disappear. That's the power of ecosystem. A specific example would be tilert , a megakernel inference engine dedicated for decode. They collaborate with vLLM by using vLLM prefill + tilert decode in , as highlighted in from @SemiAnalysis_ .

SemiAnalysis says Google's TPU beats NVIDIA by 50% per dollar. Its own dashboard, each chip at its best, says NVIDIA by 9.7x. The trick: run Blackwell with its four biggest advantages off and call it apples to apples. Free isn't cheap enough anon

Our RL stack now supports NIXL weight transfer, reducing trainer-to-inference transfer time 9x compared with NCCL: from 86 seconds down to single-digit seconds for an 800B-parameter model, and even <4 seconds in our experiments. For prime-rl users, this means over 25% more throughput end-to-end compared with our previous speed. It also clears the way for fault-tolerant, elastic inference scaling that NCCL's rigid process groups made difficult.