While serving billions of tokens, you need to watch out AWS costs too 😁
We optimised that by grouping up to 8 tokens in an LLM chunk, very efficient btw. 85% cost optimisation comes only from that 1-line config change.
364 bytes. That’s what one ~4-byte token was costing us.
@ilkh0m, @rajaguru95 and @vikashkaushik01 found it, fixed it and shipped it with no change to time to first token.
48x less bandwidth. Zero drama.
Great infrastructure is built by engineers who sweat the bytes.
1
5
106
AI infrastructure has a concentration problem.
Even neoclouds, built to challenge hyperscaler dominance, are racing to build the same kind of centralized, capital-heavy infra.
Our own @ilkh0m sat down with TechArena to break down the pricing edge neoclouds bring, the tradeoffs, and how io.net offers a real alternative.
AI agents could already spin up GPUs on io.net in under 3 minutes.
Now they can pay for them too.
Stripe + x402 support is live on io.net's Agent Cloud
Agents provision compute, top up IO Credits, and manage their own budget. Mid-session. No human required.
No pre-loaded ceilings. No hard stops. No DevOps engineer on call at 2am.
Task assigned. Result delivered.
I love reading about new product launches, but I never have enough time to keep up.
So I created t.me/launch_digest, a channel with an agent that monitors places like HN, PH, Reddit, and IH, and reposts the most interesting launches.
It solved my problem
Ilkhom Sidikov retweeted
Still paying OpenAI $10k/month to train models you'll never own?
Stop. 🖐️
@ionet Training as a Service (TaaS) is now live.
Bring any open-source LLM from Hugging Face,
fine-tune it with SFT/PPO/DPO, and own the weights. Zero DevOps. Zero vendor lock-in.
Ilkhom Sidikov retweeted
A little over a year ago, io.net was just a spark of an idea. Today, I’m blown away by what our team and community have built together.
We’ve shipped a full-blown decentralized cloud stack—Ray, Kubernetes, Bare Metal, CaaS—with VMs and Deployment APIs on the horizon. It’s now a real contender alongside the biggest players out there.
Over 300K GPUs have powered this network, with 6K+ active today and 5K+ hooked up through SuperAgg APIs. Our backbone is rock-solid and growing stronger every single day.
IO Intelligence is live, serving up top open-source models and in-house tool agents, already handling 3B+ tokens a month. RAG is up and running, and agentic workflows are next in line.
We’ve delivered $14M+ in compute, and the future is already fueled by 1K+ H200s and 800+ H100s serving enterprise-grade workloads.
This is just the beginning—step one of a thousand. Grateful for this community and what’s to come.
Let’s keep beaming forward. 🚀
Ilkhom Sidikov retweeted
Simple, fast, and endlessly flexible, @Bc8_ai_official makes it easy to bring your ideas to life—whether you’re creating art, playing with style, or crafting visuals that feel truly your own.
With an intuitive interface and powerful generation tools, you can explore, customize, and create in real time.
Built on secure infrastructure with developer-friendly tools and new features always on the way
bc8.ai/ is here for whatever you imagine next.
Ilkhom Sidikov retweeted
Resilience On Chain - Mission Report Day 1
We went from
- Saving our lives from active earthquake
- Picking up only essentials(Phones and Laptops) from the little window we were allowed after the earthquake
- Losing our homes which we built and cherished for past 3 months
Ilkhom Sidikov retweeted
We are thrilled and honored to be featured on @TechCrunch's list of unicorn startups for 2024! 🦄🚀
Check out the full list and join us in celebrating this milestone.
Thank you for the continued support! In the mean time, always DYOR & stay informed. Let's keep building! 🦾
techcrunch.com/2024/08/05/38…
#UnicornStartups #TechCrunch #ioNet











