The token economy that will power open-source AGI is here 🌐
Over the past year, we have engineered a community-first, long-term oriented system to foster open-source AI development.
Read our blog detailing the economics behind SENT 👇
teapawt.org/SentientAGI/status/201…
2
123
comolokko retweeted
At @NeurIPSConf @salahalzubi401 presented ROMA—our Recursive Open Meta Agent 🧠
ROMA’s innovative architecture results in a 45.6% accuracy on SEAL-0, the hardest SEALQA subset—a new state-of-the-art for open-source reasoning.
comolokko retweeted
Open AGI Summit is back this February at @EthereumDenver 🙌
If you’re building open-source AI infra, researching decentralized training, or believe intelligence should be built in the open, this is where you’ll want to be.
🗓 Feb 19th
📍 Denver, Colorado
The Open AGI Summit returns this February at @EthereumDenver brought by @SentientAGI.
Join the global movement shaping the future of open source AI.
Apply for your ticket below 👇
Big senti fam :)))
During @openagisummit at @NeurIPSConf, our cofounder @viswanathpramod defined one of AI’s most important use cases: High-Stakes AI.
If AI handles high value actions like payments, trades, or negotiations, it must run on systems engineered against failure and attack.
3
37
All in on @SentientAGI for 2026.
(Quote if you’re on the same page)
1
1
47
comolokko retweeted
You can throw any open-ended question at SERA and trust it to return a high quality answer in record time.
Open-source on top 😁
You don’t need to be an engineer to shape the future of open-source AI.
If you teach, translate, host, write, or build community, Sentient Sparks exists for you.
Formalize your impact. Grow your reach. Support open, transparent AI.
You do not need to be an engineer to matter in the open-source AI space.
If you can teach, host, translate, or build community, Sentient Sparks is your way to formalize that work and push accessible, open, transparent AI forward.
Apply below as a Sentient Spark and get more reach and support 👇
1
44
SERA Crypto takes that vague “what should I do?” feeling and breaks it into real, solvable risk questions.
It checks whale flows (are big players acumulating or dumping?), derivatives stress (where liquidations are hitting), and how your leverage should actually change right now.
Live on-chain liquidation data get seperated first, then recombined into one clear risk view.
Not a recap a plan.
End result:
High leverage trimmed where liquidations cluster, stops adjusted to current volatility, whale flows watched for regime shifts, part of the book de-risked so one cascade doesn’t nuke everything.
@SentientTurkiye @SentientAGI @scalpkripto @vivekkolli @shad_haq_
Reassess your crypto positions now, backed by consistent, auditable market and on-chain data.
As conditions change, SERA-Crypto turns live whale activity and derivatives metrics into a risk view you can act on, helping you adjust your portfolio in real time.
🧵 A step by step breakdown of how SERA-Crypto turns live data into a tailored risk management report.
5
9
90
Our new paper accepted to IEEE SaTML 2026 asks a simple question: are LLM fingerprints really robust in adversarial settings? Short answer: mostly no. Prior work usually assumes a benign model host, but we instead consider a malicious host and evaluate 10 different fingerprinting methods. With simple, targeted attacks, fingerprint verification can almost completely fail while keeping over 90% of the model’s original utility. We find shared weaknesses across memorization based, intrinsic, and watermark approaches: easily suppressed token spikes, unnatural fingerprint prompts, and shared statistical signals. The takeaway is clear once you assume an adversarial host, many “robust” fingerprinting schemes collapse. The path forward is fingerprints that are more natural, flexible, and harder to imitate. Open-source releases coming soon
@SentientTurkiye @SentientAGI @vivekkolli @scalpkripto
Are Robust LLM Fingerprints Adversarially Robust?
Our latest paper, accepted to IEEE SaTML 2026, analyzes the robustness of model fingerprinting under adversarial conditions and shows that simple, targeted attacks can reliably defeat many existing fingerprinting strategies.
🧵 Read more for a deep dive on our latest research
3
10
185
comolokko retweeted
In a few hours @abhishek095 will join @unstoppableweb to discuss all things .agent 🎙
Join to learn about use cases and how .agent infra lets you register agents, wire permissions, and make agents discoverable across the GRID 🌐
⏰ 11AM PT
SPACES ALERT:
@SentientAGI joins us today at 2 PM ET to talk about AI agents and the .agent TLD.
We’ll cover what’s live, what’s coming, and how builders can get involved.
Set your reminders now and join the space.
teapawt.org/i/spaces/1YqKDNypLgBJV
Open research isn’t just an ideal it’s a movement shaping the future of technology. Our mission is to build an open-source AI ecosystem powered by transparency, collaboration, and shared innovation.
All of that work has led to the GRID: a fully open, production-ready AI stack that organizations across the world are adopting to accelerate development, reduce costs, and break free from closed, restrictive systems.
The GRID represents years of open research coming together into a unified, scalable, and community driven platform. It’s more than infrastructure it’s a blueprint for an open AI economy where everyone benefits.
This is what the future of open source looks like: accessible, powerful, and built for all of us. @SentientAGI @SentientTurkiye @shad_haq_ @vivekkolli @scalpkripto
Open research has a purpose: to facilitate open-source AI through an open economy that serves us all 🌎
We’ve converged years of open research into creating the GRID—a comprehensive, production-grade open-source AI stack that enterprises are adopting every day.
This is the future of open-source 😁
3
10
220
comolokko retweeted
Looking forward to seeing everyone here to check out @SentientAGI's latest research.
Also come talk to us for sneak peeks into how we are preparing to bring all of these learnings into real production AI environments
Prof. @viswanathpramod and co-founder of @SentientAGI will present alongside a lineup of top AI leaders at the Open AGI Symposium, covering topics from AI in finance to games and agentic architectures.
Sentient researchers will also showcase one of the four NeurIPS-accepted papers — a major milestone for an early-stage AI lab.
1PM – 6PM PST.
comolokko retweeted
Our cofounder @hstyagi joined @MatthewBerman on the @ForwardFuture podcast to share his thoughts on AGI and its global development.
Complex reasoning is the next leap in AI development, and we’re excited to share some updates in our research over the next few weeks.
The first phase of AGI isn’t robots or superintelligence.
@hstyagi, founder of @SentientAGI, on what happens when the entire digital world becomes agentic.
“Our entire digital life will be AGI-fied, so we don’t have to click through everything we do on the internet.”
“We already have sophisticated review and research machines like ChatGPT that can do deep literature surveys. But our buying experience, our planning experience, everything we do online, that gets AGI-fied first.”
And the timeline is short:
“We’re very close to it, maybe two years, where the whole internet is rewired so we just express intent, and things get done, and we get the output we want.”












