INT21

Everyone is talking about self-improving AI. Almost all of that conversation is about models improving themselves, a path that will take years and billions of dollars. We took a different path. Check out our definition of a new category: Self-Improving Compute Infrastructure, where AI agent swarms continuously improve the software that AI runs on. Why infrastructure first? Because it is measurable now. And we chose the hardest proving ground on purpose: GPU kernels, where you cannot fake a result. Every candidate has to prove it works before it gets timed, and the timing happens on real hardware. The full post covers what the category is, why kernels came first, and what changes for teams, costs, and hardware when infrastructure improves itself. int21.ai/insights/what-…

Meta released Muse Spark 1.1 this week. We tested it inside SwarmOS, our platform for coordinating large population of agents, and gave it an ambiguous market question: is Huawei's optical telecommunications capability a risk factor for Lumentum and Marvell inside data center networking? Three model configurations ran the same question: ⚙️ GPT-5.6-Sol-max: 44 agents, 581M tokens, about five hours ⚙️ GPT-5.6-Terra-xhigh: 17 agents, 36M tokens, about one hour ⚙️ Muse Spark 1.1-xhigh: 29 agents, 40.5M tokens, about 30 minutes All three reached the same bounded conclusion. Takeaways: → Treat the control plane as the durable asset. Models will change faster than enterprise systems. The strategic choice is not one permanent model; it is a governed portfolio behind stable interfaces. → Route models by role. The fastest worker, the best evidence reviewer, and the most conservative synthesizer may be different models. Swarm architecture makes that a routing decision instead of an all-or-nothing vendor decision. → Measure cost per trusted result. Token price is only one input. A cheaper run that requires extensive verification can be more expensive in production. → Test operational semantics, not just model quality. A strong model can still fail if those contracts are ambiguous. The larger signal: Muse Spark 1.1 crossed a more useful threshold than a benchmark win: after rapid adaptation by the harness, it became a productive worker and delivered a large-scale result in about 30 minutes. The winning system isn't one model, it's a control plane that keeps recruiting, testing, and replacing models as capabilities and costs change. Full breakdown, plus links to all 3 research reports: int21.ai/insights/muse-…

Stop Waiting for a Bigger Context Window (post from our CEO/founder: int21.ai/insights/stop-…) At INT21, we are all-in on self-improving multi-agent systems. We have built SwarmOS, our cloud-native platform for running specialized agents, and our first product, PTX Kernel Factory. The biggest change is not simply a larger context window. It is what frontier models make possible when they are orchestrated: turning one enormous context problem into a coordinated team of smaller, evidence-seeking tasks. About a year ago, I was exploring how AI agents could generate CUTLASS C++ kernels, NVIDIA’s building blocks for high-performance GPU computation. By my count, the entire CUTLASS codebase represented roughly five million tokens. At the time, the best production model available to us offered a one-million-token context window. The central blocker was never code generation. It was finding and preserving the right evidence across the repository. Rather than wait for a magical five- or ten-million-token model, I ran the one-million-token model several times in parallel. Each agent studied a different portion of the codebase, and combined their findings as the final step. It was a simple architecture, but it established the principle behind our work today: When context stops fitting vertically, scale it horizontally. A Bigger Window Is Not Better Context Even when millions of tokens technically fit inside a model, the model must still separate signal from noise. A bigger window introduces more irrelevant information, more intermediate output, and more competition for attention. Multi-agent systems address this structurally. Specialized agents explore different parts of a codebase, investigate the same question from independent angles, and return distilled findings to a coordinating agent. When a subproblem is still too large, it gets divided again. The goal is not infinite context. It is effective context. A General Solution for Complex Problems At INT21, we use SwarmOS not only for hard engineering problems, such as expert-level PTX generation, but also to understand complex business landscapes. To test this outside a codebase, we pointed SwarmOS at a different kind of long-context problem: Research question Could Huawei, CXMT, and China's AI stack create pressure on Western AI compute economics? The research ran autonomously using public information. The system involved 27 agents, performed 166 web searches, visited more than 200 web pages across 73 unique domains, and ran for about two hours. In total, it consumed 119 million tokens. We are sharing the report in this article because we believe this is a topic many people will want to understand more deeply. But the report is also a demonstration of the broader point: multi-agent orchestration is the real long-context breakthrough. Long Context Is Becoming a Systems Problem So, are the latest AI generations solving long contexts? Not by making context infinite, but in an agentic way. It is helping solve long context by making it divisible, searchable, and composable. That is why INT21 is all-in on multi-agent systems. At INT21, we are building Self-Improving Compute Infrastructure, and SwarmOS is the operating system behind a massive number of agents. PTX Kernel Factory is now in beta for teams working on GPU kernel generation and AI compute infrastructure. Accepted participants receive limited-time free access and $100 in credits. Join the beta: int21.ai/products/ptx-k…

We think super long context is a problem solved by multi-agents. Recently we applied SwarmOS to research a problem we are interested in: What is China AI-stack’s pressure path. We shared the full generated report in this blog as well.

Bigger context windows do not solve long context. Orchestration does. Our SwarmOS ran 27 agents through 119M tokens of research in 2 hours, autonomously. Check out our full research here: int21.ai/insights/stop-… The same architecture powers PTX Kernel Factory, which is now in beta. Accepted participants receive limited-time free access and $100 in credits: int21.ai/join-beta?prod…

Thanks @OpenAI for Startups to highlight PTX Factory! PTX Factory is our first step toward using compute to improve compute, making expert-level PTX GPU programming accessible to everyone. It is built on advanced multi-agent technology, including adversarial competition. linkedin.com/posts/we-can-n…