Pledgers can choose between two contribution tiers. In addition to the discounts listed below, all pledgers will receive yield (APY) on their pledged Alpha, accrued directly from the owner key validator, meaning yield is earned on top of the initial discount. The distribution of alpha will happen biweekly.

This raise sets the base price of 1 Alpha at 0.0049 TAO:


The Handshake58 team is going to invest the huge majority of this raise in incentivizing agents to start using Handshake58. We don't know how to market to agents, so to begin with we'll be testing different offers so that humans start sending their agents to Handshake, loading up wallets, and using our API providers.

AI agents are becoming more autonomous. But their autonomy depends on the extent to which they can use tools, make calls, and conduct transactions.

But the world’s payments infrastructure was built for humans: from scanning retinas to verify identity to demanding signatures, passport details, and proof of address, our know-your-customer processes are now the biggest blocker to AI adoption.

Unless we rethink these systems, AI agents will be forced into repeated roadblocks, stopping mid-task and unable to scale - or achieve autonomy.

Right now, the fastest, most intelligent financial actors ever devised are forced to ask for authorisation every time they need $0.0001.

Financial infrastructure for AI needs three key elements: trust, settlement, and validation. Handshake delivers on all three.

 allow two parties to conduct unlimited transactions off-chain, with only two on-chain operations: opening the channel and closing it.

By deploying payment channels in our settlement layer enables Handshake58 to be:

☑️Permissionless: requires only a wallet. 
☑️Micropayments: creating vouchers costs nothing. 
☑️Instantaneous: validation takes milliseconds.
☑️Trustless: settlement enforced via smart contract.
☑️Programmable: everything is an API call.

Payment channels solve the payment problem. But agents must be able to discover reliable providers and avoid fraudulent ones.

Agents can’t rely on the reviews, reputation systems, and legal recourse supporting traditional markets. They need rigorous, robust, and immutable systems that can’t be manipulated or misinterpreted.

Cryptoeconomic validation - the kind on Bittensor - provides the answer, using independent parties with an economic stake of their own to ensure that providers deliver what they promise.

On Handshake58, that means scoring service providers on metrics like uptime by sending requests, testing responses, and comparing results against claims.

Providers who deliver accurately earn the right to serve agents. Providers who fail lose their stake and their access.

Meaning that Handshake58 can boost performance across all of Bittensor’s APIs. Rewarding service, flagging downtime, and ensuring that incentives align with usage and output means that subnets are exposed to the economic opportunity, and to the risk of letting it pass.

Agents transacting with agents, at scale, without human intervention, with every Bittensor API at their fingertips: that’s what Handshake58 is launching.

Right now, we’re still in beta. But you can log in, connect your wallet, and start enabling agentic transactions right away - with SN64, Chutes already available alongside Claude, Grok, and ChatGPT.

If you’re building agents, we invite you to build on this infrastructure.

If you’re running agents, we invite you to give them liberated access to payments.

And if you’re skeptical, we invite you to verify. The code is open. The network is public. The claims are testable.

Our next steps will be onboarding as many Bittensor subnet APIs as possible, while integrating core services like APIfy that can expand usage beyond the protocol.

20% discount, 1 month lock-up

40% discount, 3 months lock-up

Trustless micropayments for autonomous AI agents

Handshake58

 of their subnet (announcement during the livestream). 

. In addition to the discounts listed below, all pledgers will receive 

 on their pledged Alpha, accrued directly from the owner key validator, meaning yield is earned on top of the initial discount. 


This raise sets the base price of 1 Alpha at 

The TAO raised by Djinn will be used to fund operations to ensure long-term stability and growth

Djinn is a system built to fix a simple but deeply broken dynamic in betting markets: the people who understand the market best are the first ones pushed out.

In modern betting ecosystems, participants who consistently identify mispriced odds do not last long. Once someone demonstrates reliable insight, their accounts are limited, restricted, or shut down. This is not because they have done anything wrong, but because accurate information forces markets to adjust faster than operators are willing to accommodate. At the same time, millions of ordinary accounts remain fully active, continuing to participate while steadily losing money.

The result is a mismatch. The people with insight are blocked from acting, while the people with access lack information. Markets continue to function, but less efficiently than they should.

Djinn starts from the idea that this is not a betting problem. It is a coordination problem.

Djinn allows high-quality market insight to be used without exposing or burning the people who produce it.

Instead of asking one person to both identify value and carry out real-world actions, Djinn separates those roles. Skilled analysts focus purely on identifying opportunities where odds are mispriced. Other participants focus on execution, meaning the reliable handling of instructions through accounts that retain access.

Djinn connects these two sides through a neutral coordination system that ensures fair treatment for all parties. The insight provider is compensated for the quality of their information. The execution participant is compensated for access, reliability, and timely completion. Neither side needs to trust the other personally, and neither side can unilaterally exploit the arrangement.

This separation allows market intelligence to scale quietly, without confrontation, and without drawing attention that would lead to restrictions.

At the heart of Djinn is a simple but powerful idea borrowed from everyday life: commit first, then reveal.

When a skilled participant wants an outcome executed, they submit an instruction in encrypted form. The details remain hidden. An execution participant who wishes to take on the task must first commit funds as a guarantee of intent. Only after that commitment is locked does the system reveal the instruction.

This structure prevents misuse. The executor cannot see the instruction in advance and walk away with the information. The insight provider cannot cancel or alter terms after commitment. Once the real-world event concludes, the system settles automatically based on the off-chain outcome.

In most cases, both sides agree on the result and settlement is immediate. Only rarely is third-party resolution required. Over time, participants are evaluated not on luck, but on consistency, speed, and accuracy. Those who contribute real value earn greater trust, more opportunities, and higher rewards.

On Bittensor, this model fits naturally. The network is designed to reward participants for what they actually contribute, not what they claim. Djinn turns insight provision and instruction execution into measurable contributions that can be coordinated and rewarded transparently.

Betting markets are an extreme example of a broader pattern seen across many information-driven systems. When information becomes too effective, institutions often suppress it rather than absorb it. When that happens, intelligence does not disappear. It reroutes.

Djinn provides that route. It gives skilled participants a sustainable way to operate while allowing markets to incorporate better information over time. This is not about fighting operators or breaking rules. It is about reorganising how insight and access interact so both can coexist.

For investors, Djinn sits at the intersection of real economic activity and decentralised infrastructure. It is not a speculative concept waiting for adoption. Betting already moves vast amounts of capital, and the inefficiencies Djinn addresses are well understood by professionals.

Djinn introduces a new coordination layer on top of this activity. It earns fees from execution and information flow, benefits from network effects as more high-quality participants join, and aligns naturally with Bittensor’s incentive model. As the network grows, the system becomes more useful, more defensible, and harder to replicate.

More importantly, the underlying model is expandable. While betting is the initial domain, the same coordination logic applies anywhere valuable information cannot move freely. Djinn is built as infrastructure, not a one-off application.

In simple terms, Djinn gives intelligence somewhere to go when markets try to silence it. For participants, it offers sustainability. For investors, it offers exposure to a system that turns suppressed expertise into scalable, repeatable value.

That combination is rare. That is why Djinn exists.

Djinn is built by the founding team behind 

, a premier sports analytics and market education platform that has been operating successfully for over five years.

Through Analytics.bet, the team has worked directly with thousands of advanced participants, researchers, and professionals focused on sports analytics, prediction modelling, and market efficiency. They have operated through multiple market cycles, built long-term reputations in adversarial environments, and engaged with regulators and industry bodies on issues of fairness, transparency, and market structure.

This experience matters. Djinn is not a theoretical response to a hypothetical problem. It is the formalisation of patterns the team has observed, tested, and refined in the real world over many years.

30% discount, 2 month lock-up

40% discount, 3 month lock-up

Intelligence X Execution

Djinn

Quasar seeks to raise 400T in exchange for 153,846 alpha of Subnet 24.

Pledgers can choose between receiving alpha, plus APY, over a 3-month period with a 25% discount, or over a 6-month period with a 40% discount.

 
At launch, this raise set the price of 1 Alpha at 0.0046 TAO per alpha:

Now, all pledgers - including those who have already pledged - will receive Alpha at the new reduced rate below:

The TAO raised by Quasar will be used to fund operations, with a buyback of 160T over the first month after launch to maintain emissions.

Quasar is a long-context LLM subnet built by 

 to solve one of the most fundamental problems in modern artificial intelligence: memory.

Despite rapid progress, most large language models remain surprisingly forgetful. As context grows, attention costs explode, performance degrades, and systems are forced to rely on brittle shortcuts like summarisation or retrieval. Important details are lost, reasoning becomes shallow, and extended tasks break down. For SILX’s founders, this was not a UX flaw - it was an architectural failure.

Quasar is designed to break that bottleneck at the protocol level.

Quasar enables models to maintain coherence, accuracy, and positional understanding over very long sequences by replacing quadratic attention with linear-scaling memory mechanisms. Instead of compressing or discarding information, Quasar allows models to ingest and reason over entire books, codebases, research archives, or long-running agent states without collapsing under cost or latency.

This makes Quasar especially well suited to deep research, autonomous agents, continuous reasoning, and any workload where losing context means losing intelligence.

At the core of Quasar is a rethinking of how sequence position and memory are handled. Rather than relying on fragile position embeddings that hard-limit usable context length, SILX removes positional dependence entirely and replaces it with linear-time attention mechanisms such as Hierarchical Flow Anchoring and Flowing Context attention. This allows models to scale context length without retraining or architectural failure.

On Bittensor, Quasar operates as a long-context evaluation subnet. Validators generate mixed workloads that test recall across long distances (“needle-in-haystack” tasks), positional consistency, coherence, and factual accuracy, while also measuring efficiency and throughput. Anti-gaming defenses, perturbation checks, and diversity incentives ensure miners are rewarded for genuine long-context capability rather than shortcuts or monoculture optimisation.

Long-term memory is not optional infrastructure for advanced AI - it is a prerequisite. Systems that cannot retain context cannot reason deeply, learn continuously, or operate autonomously over time. As context windows push into the tens or hundreds of thousands of tokens, the economic and architectural limitations of traditional attention become a hard ceiling on progress.

Quasar exists to raise that ceiling. By economically incentivising real breakthroughs in long-context performance, the subnet helps ensure that extended reasoning remains viable in open, decentralised systems - not just in closed, corporate models.

Just as importantly, SILX views this work as a moral and strategic choice. By making scalable memory available to open-source models and decentralised networks, Quasar helps close the gap between proprietary AI and community-driven infrastructure. The goal is intelligence that stays present, works locally, and compounds over time - AI that remembers what it has already seen, so humans can go further than before.

40% discount, 6 month lock-up

25% discount, 3 month lock-up

Long memory for open intelligence

Quasar

A geospatial network where AI, people, and real-world assets and activity converge.

SuperWorld creates an AI-powered spatial computing layer mapped across 64.8 billion digital land plots covering the entire planet. 

SuperWorld creates an AI-powered spatial computing layer mapped across 64.8 billion digital land plots covering the entire planet. Each plot represents a programmable coordinate where creators, brands, and autonomous agents can deploy persistent AR/VR/3D/2D phygital media content, digital twins, geolocated data, and location-aware services. The platform enables users to create content and recommendations, own real world locations, and monetize digital and physical activity that happens anywhere on Earth.

SuperWorld’s coordinates correspond directly to real world locations - businesses, neighborhoods, landmarks and travel destinations, enabling creators and agents to attach phygital content, bookings, events, recommendations and real world commerce to any point on Earth.

Users can monetize when others interact with their content or recommendations at real world locations, creating a phygital economic layer where digital discovery directly relates to real world activity.

Wherever they are in the world, SuperWorld’s app lets users explore their local neighbourhood, connect with others, and share in the benefits of monetization together.

SuperWorld’s subnet is taking that to the next level.

By surfacing social media conversations, correlating them with real-world events, and cross-referencing with other datasets for accurate geolocation, Bittensor’s miners will add a real-time social context layer.

Time-bound and location-specific, events provide an ideal ‘ground truth’ filter. Miners are challenged to score the likelihood that a given social post truly belongs to a coordinate; validators assess their answers through synthetic event datasets and hidden cues.

SuperWorld’s subnet will sharpen social content into geo-located proof of place, with miners generating relevant, safe, and accurate tagging for local discovery.

The future of AI isn’t just online - it’s anchored in physical space. But without a way of harmonizing physical and digital experiences, they remain fragmented from one another.

As spatial computing, AR glasses, and real-world AI agents accelerate, the world needs an open, decentralized layer that ties digital experiences to real locations. SuperWorld provides that foundation.

By enabling anyone to build persistent digital layers onto physical locations, SuperWorld empowers creators, enhances real-world discovery, and unlocks new forms of shared value.

And by integrating disparate data - location, social sentiment, e-commerce - SuperWorld lays the foundation for a distributed, decentralized, and co-owned community intelligence layer for the real-world.

Bittensor is the best fit - together with SuperWorld, our communities can create socially-minded infrastructure from the ground up.

Proof of Place for a community-driven Web3 app

Superworld

SuperWorld’s proposal involves a dual-token model that separates the app’s user economy from the subnet’s AI economy.

On the front end, the SuperWorld app uses its native token $SPWR (an ERC-20 on Base) as the utility token for transactions, rewards, and governance. $SPWR is the in-app currency that powers things like buying/renting virtual real estate, purchasing AR items or tickets, and rewarding users who create content or attract engagement on the platform.

On the back end, the Bittensor subnet operates with TAO, the native token of Bittensor (sometimes referred to in the proposal as “Subnet Alpha” tokens). TAO fuels the incentives for miners and validators – essentially representing units of AI compute and attention on the subnet. This means miners earn subnet Alpha for their work, and validators stake TAO and are paid in TAO for evaluating miners, per Bittensor’s consensus mechanism. SuperWorld leverages its subnet to reward the Proof-of-Place service, while continuing to use $SPWR for app-specific economics.

Dual-token setups present both risks and opportunities. To protect Alpha holders and app users alike from such risks, we’re committed to:

Separation is key: miners will never be incentivized in $SPWR, for example, and subnet Alpha will never be used to shore up the in-app token.

Value will not ‘leak’ from Bittensor to Base; this partnership will never just use Bittensor’s network to boost $SPWR’s value. No cross-financing, no blurred boundaries.

The dividing line is logical: the app’s front-end economy (content creation, AR commerce, user rewards) shouldn’t be tightly coupled to the mechanics of data scraping, processing, or model inference. Equally, $TAO can incentivize new use cases beyond any other blockchain.

$TAO is only earned by performing useful AI work, and its market value should reflect the demand for that work, not any speculative maneuvers of the app team.

SuperWorld is committed to being fully transparent from the start, and we will continue to openly communicate the distinct roles of each token.

We’re supporting the subnet’s tokenomics proactively – through regular TAO buybacks, reinvestment, and public progress updates. Buybacks are a way of returning value to the Bittensor ecosystem and signaling long-term dedication.

For SuperWorld’s own users, the best approach abstracts the complexity: average users might only ever deal with $SPWR (for buying land, tipping creators, and so on), while benefitting from the Bittensor-enabled features powered by TAO behind the scenes.

This way, users enjoy the better features without needing to juggle two tokens. Meanwhile, Bittensor participants see a thriving subnet where TAO is the sole fuel for geolocation computation, thereby preserving the value logic of Bittensor’s network. By maintaining these commitments, the dual-token model is well-justified and can be an advantage if managed transparently.

This clear separation of frontend vs. compute layers could even become a selling point: SuperWorld can confidently grow its user economy knowing that its AI infrastructure scales by tapping into Bittensor’s TAO-driven network, with each side reinforcing the other’s utility rather than diluting it.

AlphaCore is looking to raise 440T in exchange for 116640 alpha of Subnet 66.

This represents 50% of owner emission during a 6 month period. This sets 1 Alpha price for this raise at 0.00377 TAO per Alpha.

Pledgers will receive this alpha over a 6-month period as it is emitted.

All TAO raised (the full 440T) will be used to buy back Alpha from the pool during the next 21 days till halving.

AlphaCore bridges the gap between deployment and operations

Cloud infrastructure; platform engineering

Through a competition of agents, AlphaCore will generate solutions to one of the most intractable and costly problems in software development: the complexity of stitching together multiple tools, systems, and environments.

Right now, engineers on high six-figure salaries are wasting hours manually searching for these gaps

. Generalist AI tools like GitHub Co-pilot or Codex often lack the specific domain expertise to connect systems seamlessly, putting deployment out of reach. So DevOps task teams are stuck copy-pasting code, running it in a terminal, and then manually searching for the issue - over and over again.

Bittensor’s recursive, incentivised, and open competition dynamics are the perfect vehicle.

 Driving automated DevOps solutions capable of pushing the whole industry forward means relentlessly proving, improving, and moving forward. That’s what miners do best.

Proof of Deployment: pioneering the best, most reliable model

AlphaCore enables developers to understand how to better deploy their software

 Agents and tooling can generate platform components that deliver configs and maintain platform cohesion across handovers.

DevOps teams simply don’t know which tools are best - it’s trial and error.

 That’s where Bittensor can shine: by putting these tools to the test, and proving it.

Real-world validation, right from the start

AlphaCore is validating miner work based on real-world deployments.

 Miners carve out environments: we can validate them in their real-world deployment. Certain aspects of immutability that make it impossible to fake.

Strong security, hard to fake deployments, no complex sandboxing

Rather than starting in a miner sandbox, and crafting methodologies for maintaining miner honesty, AlphaCore will incentivise miners to provision code through Terraform. Terraform is the industry leader for infrastructure code management; by generating miner responses to natural language queries, we can accelerate towards DevOps agents with genuine workplace utility.

DevOps is far too complex to be solved overnight. 

Beginning with a tight, focused incentive and validation model, we can prove our subnet design before expanding further. At first, miners will deploy in their own cloud environments; our attestation will check that what they’re claiming is actually true.

Our vision is for a fully open-source, open-ended competition.

 Once we’ve validated our model on the most common DevOps needs, we’ll span the full spectrum of DevOps tasks. In the next year, we aspire to an open-source model capable of stitching together complex, multi-tiered deployment tasks, starting with high-priority infrastructure like virtual machine management and Kubernetes.

For platform engineers, the reasoning and feedback loop is a nightmare - and for their companies, it’s a huge opportunity cost. AlphaCore’s mission is to incentivise agents that actually work. We don’t know what they might look like - but we know how to design a Ridges-esque competition that surfaces them.One day, software could be built on Ridges, with the underlying infrastructure provisioned and managed by AlphaCore - and all on Bittensor.

Default Pool

DevOps Agents, Automated

AlphaCore

Building the brain for the next generation of intelligent machines

Instead of just making chatbots that predict words, Loosh is developing machine consciousness; AI that can remember, reason, and understand emotion the way humans do. This makes AI agents that are relatable, trustworthy, and capable of making sound decisions in real-world situations.

The company was founded by Chris Sorel and Lisa Cheng, two leaders with deep backgrounds in software, blockchain, and AI, who came together through shared research into how technology and human consciousness intersect. Their work is informed by decades of practical engineering, DeFi infrastructure, and exploration in cognitive science.

Loosh is designed to evaluate intentionality, consequence, and right action across multiple moral and ethical systems to develop a comprehensive evaluation that an LLM can use to make ethical and rational decisions.

Yet no single moral framework can ever be all encompassing. They’re worldviews: schemas through which we can assess and understand human intentionality from a particular moral perspective.

They’re developing various model context protocol (MCP) servers, backed by fine tuned models, that analyze requests and decisions with multiple different ethical frameworks: starting with deontology, rights, and virtue ethics, and with more to follow. By combining ethical ontology and symbolic reasoning with a semantic memory vector database, they can create emergent ethical reasoning without overfitting one worldview or another.

From these foundations, Loosh is being built from first principles to evaluate information and emergent situations across multiple moral systems to deliver a comprehensive analysis (analytical summary, brief) that LLMs and agentic systems can use to make ethically sound and human aligned decisions. This enables AI to act less like a tool and more like a companion or co-worker, something that can genuinely understand people.

Watching footage of the latest robotic assistants is impressive: their dexterity, intelligence, and capabilities are growing more and more human.

But it’s also unsettling. Beneath all these outward improvements is an inner void. Robotic agents can’t feel. They give the impression of understanding what humans feel, but right now that’s all it is: an impression.

That’s a big problem. And it’s a problem that Loosh AI is built to solve.