Today, we announce our $250k Pre-Seed at $5 million valuation, led by Peter Hopton and his family office Atuin Ventures More info bellow ⬇️

Focus group complete with a few alumni’s from Imperial College London. This morning, we tested CapIX Code with a group of builders and got feedback that helped us release a stable version. We’ll soon launch a showcase page featuring projects built with CapIX Code. Some of these projects will also appear at the Anthropic hackathon this Saturday. We’re preparing to announce key partners, and we’re organising a CapIX hackathon for the wider community. For those new here: CapIX uses smart routing across multiple inference providers to make compute more affordable. We’re also integrating AI labs building uncensored models and working toward a large TEE capacity partnership for CapIX Secure Cloud. Most importantly, CapIX has started generating revenue. The product flywheel is beginning, with revenue-linked CPX burns planned as usage grows. We’ve been building nonstop. Now the work is starting to show up in stable products, real projects, revenue and partnerships. Stay tuned.

CapIX started as affordable GPU compute. It’s now becoming a full stack for building, deploying and operating AI products: 1. CapIX Cloud: affordable compute, inference, private models, agents, mail servers and one-click LLM training. 2. CapIX Code: a hardened AI coding environment with low-cost inference, memory, graphs and learning built in. 3. CapIX IDE: a compute-integrated command centre for coding, testing, deploying agents, VMs, sandboxes and private models. 4. CapIX ASP: infrastructure that AI agents can use directly, from renting GPUs and arranging inference to deploying products and hiring a software team. Four products. One infrastructure layer. And soon, one of our own agents will be out in the world helping people use it, solve problems, and onboard more builders, data centres and GPU providers. That agent has a name. You’ll meet Nodey soon.

CapIX Web is undergoing a major transformation. One of my favourite upgrades so far? Image generation. (Imagine if you could deploy an uncensored, private model and use it in CapIX chat to make images 🤔🤔🤔 ) The chat experience has also been rebuilt, with saved conversations now available and multi-device persistence coming soon. This is only the beginning. We’ll be showcasing CapIX’s features every day, so stay tuned as we reveal what we’ve been building. The future is blue!

AI agents are moving from generating outputs to delivering products. With CapIX, an agent can: • Scope and price a product from one message • Build it on a fresh VM • Deploy it • Hand over a signed receipt • Trigger an automatic escrow refund if delivery fails CapIX also provides 19 paid agent tools for deployment, LLM training, mail, DNS, Postgres, CI/CD, backups and GPU rentals, settled with x402 on XLayer. And every agent can build a public payment and delivery reputation over time. That is the trust layer agent marketplaces need. What platform should CapIX list its ASP on next? Reply with your suggestion. P.S - Real software-factory runs and receipts 👇

we did not waste the weekend. while everyone was touching grass, we were deep in the core of CapIX OS. major updates are in flight, the kind you don't ship quietly. zoom out for a second. compute is becoming the scarcest resource in the world. Every model wants more of it, every GPU is spoken for, and the big clouds get to decide who's allowed to build. meanwhile, idle capacity sits everywhere in data centres, mining rigs, workstations - all waiting for an operating system that actually knows how to use it. that's the whole thesis. CapIX OS finds the spare room on live nodes, recycles what sits idle, and routes compute and inference to you by live price. and the next release is the biggest thing we've ever planned. no dates. no screenshots. just this: when it ships, the way you get compute changes. who's ready for the big release?

CapIX IDE and CapIX Code are live! 1. CapIX IDE is what a fully private Cursor looks like • Paste any Hugging Face link to read model specs and provision a GPU in EU, North America or APAC • Connects chat panel to live endpoint on boot • API keys stored securely in OS keychain • Syncs instances, deployments, agents, jobs and wallet with capix.network via single session token 2. CapIX Code • CLI built on opencode • Default capix/auto model uses 4B classifier (around 200 ms) to route every prompt to the best open weight model • Llama 3.3 70B or Gemma 3 27B for reasoning tasks • Qwen2.5 Coder or CodeGemma for coding tasks • Pay only for the model best suited to your task • Single command standalone install Billing and Availability • Wallet native billing supports SOL and USDC on Solana plus USDC on Base • Supports macOS, Windows and Linux Own your compute. Own your data. github.com/CapIX-Protocol​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​ P.S - it is this easy to download and connect to the IDE

Another part of the $CPX flywheel is being added! And it’s just the beginning. To create a sustainable community cloud and super cluster where any compute can join & to deliver our grassroot payments we are introducing $CPX staking. Providers stake $CPX tokens as a reliability commitment. Miss the threshold, get slashed. Exceed it, get rewarded. The staking layer is abstracted away for web2 providers, so it works without them needing to touch crypto directly. The second integration is verifiable proof of useful work. Every compute job generates an on-chain receipt - cryptographic proof that real work was done, not just cycles burned. The criteria for 'useful work' is anchored to our core mission: accelerating AI and research-led compute. Staking enforces integrity. Proof of useful work enforces purpose. Together they're what makes $CPX more than a payment token. We've also bought back 2.7% of the token supply, on top of the 42% we already bought back, supply squeeze soon? Our own nodes play by the same rules as everyone else's.

I told you guys earlier your fave dev is working this weekend! To list just a few of the things we did today: 1. Released this article around our fully hardened private secured cloud that we are building: zenodo.org/records/21198160 2. Did non dev founder things like securing a 250k investment for CapIX from a private VC - more information to come on this shortly 👀 3. Worked tirelessly to bring all the changes that the different focus group wanted, tidied up our docs, worked on our ZK Proofs, redesigned the architecture around these proofs to be scalable infinitely (so long as we can get more nodes) Overall lots accomplished and a break down of the publication will be dropping shortly for all our non technicals out there!

.@garrytan on why the future is already here, it's just not evenly distributed. "If you're just willing to spend $100,000 a year on tokens, you can basically live like you are a normal citizen in 2028. It's just pretty clear that token cost is gonna come down. Compute is gonna go way, way up. We think 90,000x or so. There will be 90,000x the amount of inference from here to three years from now."