Anyone can build AI. Few can make it truly work. HOOTi is the browser for real AI — powered by shared compute and MAGNET. Built by Holo Studio Co., Ltd.
Seoul, South Korea
Joined February 2026
- Tweets81
- Following180
- Followers39
- Likes92
Pinned Tweet
HOOTi Update: Fairer AI Evaluation, Better Research Experience
This release improves how MAGNET evaluates AI-generated research, making comparisons more transparent, reports more readable, and verification more reliable.
From fairer head-to-head judging to a redesigned research report and more robust IPFS retrieval, we're continuing to build trustworthy AI infrastructure.
Thread 🧵
TGE is coming for @StockClaw
The big boss is showing 🌟
This Post is from a suspended account. Learn more
[HOOTi Update Announcement] - Major update shipped
We've fundamentally reworked the network's reward accounting (CV20).
Reward issuance is now unified through a single path. Every reward pool is always fully backed, so paying out beyond its backing is structurally impossible.
All 10 validators upgraded to the new version with zero downtime.
Live and running normally, no service impact.
2/ Better Reports, Stronger Infrastructure
This update also introduces a dedicated report experience for completed research.
Instead of reusing the model interface, synthesis results are now presented as structured research documents with:
• A document-first layout
• Grounding and verification panels
• Hypothesis cards
• Full bibliography
• Transparent head-to-head comparisons
Behind the scenes, we've also improved baseline reliability and IPFS retrieval.
Legacy datasets can now be fetched through multiple gateways instead of relying on a single cached source, making historical research far more resilient.
Hooti retweeted
@HoloStudioKR team have been quietly working on something bigger than individual products.
Introducing the first MVP of the Holo Ecosystem, a unified portal where AI, Gaming, Finance, Infrastructure, and Social applications come together in one place.
Instead of scattered products, we're building an ecosystem where each app strengthens the others. Current featured apps include 9 apps across 6 categories and this is only the beginning.
Next, we'll continue expanding the ecosystem with:
→ More applications
→ Better discovery
→ Cross-product integrations
→ Developer & partner onboarding
→ A connected experience across every Holo product
More updates soon. 🚀
📢 HOOTi / MAGNET Update
A genuinely good model should never be discarded because of a technical mismatch.
Over the past two weeks, we've upgraded MAGNET's model verification system to ensure leaderboard scores are based on independently reproduced results.
The update is now live on Devnet. More in Thread 🧵
1/ How MAGNET Verifies Models
MAGNET has never been designed to trust self-reported benchmark scores.
When a model is submitted, independent validator nodes replay it on a holdout dataset and reproduce the results before reaching consensus on an official score.
The network records what validators can independently verify, not what the model producer claims.
This update significantly strengthens that verification process by improving consensus around replayed results and making it easier to detect inflated, inconsistent, or overfitted submissions.
In short: Leaderboard rankings are earned through independent validation, not self-reporting.
2/ What's Changed
Previously, if a submitted score differed from the independently reproduced result, the entire job could fail as verification_failed, even when validator nodes agreed the model itself was genuinely good.
Now, the network behaves differently.
If validators reach consensus on the reproduced score, that verified result becomes the official record and the job completes successfully.
A recent example illustrates the improvement:
• Self-reported score: 73.4%
• Independently verified score: 87.3%
Under the previous system, this model would have been discarded.
Today, MAGNET records the independently verified 87.3% as the official score.
This upgrade was deployed through governance across all 10 validator nodes and builds on our ongoing work around distributed storage, reproducibility, redundancy, and network reliability.
The update is now live on Devnet.
Evidence:
magnet.hooti.ai/runs/job-d9e…
Hooti retweeted
The best of tech and crypto, combined. 🚀
Trade Tesla, Circle, SanDisk, Micron, and Nvidia 24/7 on @BNBCHAIN !
1:1 Backed: Every bStock matches a real share.
Auto-Compounding: Dividends are automatically reinvested.
Video via our AI partner @playartsdotai
AI Sovereignty Is No Longer a Theoretical Question
The recent suspension of access to Fable 5 and Mythos 5 has sparked significant discussion across the AI community.
Regardless of one's position on export controls, national security, or regulatory policy, the event highlights a broader reality:
Advanced AI models are becoming critical infrastructure.
When access to a frontier model can be altered, restricted, or withdrawn through a single decision, the conversation extends beyond a specific company or product. It becomes a question of resilience, access, and sovereignty.
The AI industry has largely focused on scaling models, improving benchmarks, and expanding capabilities. Far less attention has been given to what happens when access itself becomes a point of dependency.
History suggests that foundational technologies become most impactful when innovation is distributed across many participants rather than concentrated within a small number of institutions.
The internet succeeded because no single organization controlled access to information.
Open-source software accelerated progress because anyone could inspect, improve, and build upon it.
AI will eventually face the same challenge.
Safety, security, and responsible deployment remain essential. These principles should not be compromised.
At the same time, resilience requires open research, transparent systems, distributed infrastructure, and broader participation in the creation of intelligence.
📍The long-term question is not whether safeguards should exist.
The question is whether the future of intelligence should depend on a handful of centralized points of control.
At @HootiBrowser, we believe the next generation of AI infrastructure should be built around verifiable research, community participation, distributed model development, and open knowledge creation.
Not because decentralization is an ideology // Because resilience is a requirement.
The future of AI should not only be powerful 👉 It should be sovereign.
@AnthropicAI @OpenAI @GoogleDeepMind @MetaAI @PrimeIntellect @NousResearch
The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees.
The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance.
Access to all other Claude models is not affected.
We apologize for this disruption to our customers. We believe this is a misunderstanding and are working to restore access as soon as possible.
Read our full statement: anthropic.com/news/fable-myt…
📚 Vault & Library Are Now Live on HOOTI MAGNET
Research should not disappear after a single run. Explore:
📍 magnet.hooti.ai/vault-search
📍 magnet.hooti.ai/library
Vault and Library provide a new way to organize, discover, and reuse datasets generated through the @HootiBrowser network.
Every research topic can now contribute to a growing knowledge layer for future AI systems.
More in Thread 🧵
The datasets generated through HOOTI Research are now organized like an online library.
Over time, these datasets may power:
• Specialized LLMs
• Domain-specific AI models
• Future image generation systems
• Autonomous research workflows
• New incentive systems for dataset contributors
We believe datasets are becoming one of the most important assets in the AI economy.
Not just for training models today, but for enabling entirely new forms of intelligence tomorrow.
Vault and Library are our first step toward making that knowledge layer accessible to the community.
*Note: The features are still in an early stage and not yet fully stable, but they are now open for everyone to explore.
A more stable release is coming soon, powered by @HoloStudioKR
Hooti retweeted
The future of decentralized AI is not only open models.
The real breakthrough is open compute + open models — where knowledge and computation are both verifiable and free from central control.
This is why we build HOOTI.
@hootibrowser @opentensor
This is exactly why we built HOOTi.
AI is changing how people create, learn, and work. But the real unlock is giving everyone a simple, social, and creative way to use it every day.
The next generation of AI-native creation needs its own playground.
The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees.
The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance.
Access to all other Claude models is not affected.
We apologize for this disruption to our customers. We believe this is a misunderstanding and are working to restore access as soon as possible.
Read our full statement: anthropic.com/news/fable-myt…
📢 HOOTI MAGNET Network Update
A major infrastructure milestone has been completed.
✅ 10 seed nodes fully operational
✅ Persistent SSD storage deployed across the network
✅ Parallel multi-user research now live
✅ Zero downtime rollout
🧠Upcoming: rc57 Upgrade (Estimate June 13)
→ Multi-node DiLoCo Training
→ Security upgrades
→ Faster governance
→ Improved node reliability
We're moving from distributed research toward distributed model training.
More in Thread 🧵
1/ Studio Context
Over the past week, we completed a significant upgrade to the @HootiBrowser network infrastructure.
Seed Nodes 2–10 now operate with dedicated 200GiB persistent SSD storage, allowing datasets, checkpoints, and training artifacts to survive restarts and enabling larger-scale distributed training workloads.
We also introduced dedicated node placement with anti-affinity, improved health monitoring, synchronization behavior, and resource allocation.
The entire rollout was completed without chain downtime, and all 10 seed nodes are now fully synchronized.
2/ More importantly, the upcoming rc57 upgrade activates on June 13.
This introduces v9 DiLoCo multi-node training, allowing multiple nodes to collaborate on a single training round through shard-based learning and quorum merging.
In addition, we've deployed:
• Claude tier self-healing
• C1 signature verification security upgrades
• Governance acceleration (24h standard voting + 4h expedited path)
• Reduced capability probe overhead
Today, multiple users can already submit research topics simultaneously, with the network automatically distributing work across available nodes.
Following rc57, multiple nodes will be able to train together on a single model.
In summary: Build a decentralized network where research, datasets, training, and intelligence are created collectively by the community rather than controlled by a single organization.