# Intelligent Internet

> A planned sovereign AI protocol with a three-layer blockchain (Bitcoin Core fork), dual-currency economy, and Proof-of-Benefit consensus. Founded by ex-Stability AI CEO Emad Mostaque. Pre-launch.

- Scores: Freedom 4.2/10 (grade D)
- Category: platform · Token: FC
- Published: 2026-03-02 · Updated: 2026-10-04
- Canonical: https://ownyourmind.ai/projects/intelligent-internet/
- Data (scores + integrity verdicts, JSON): https://ownyourmind.ai/api/projects/intelligent-internet.json
- Cite as: Own Your Mind (ownyourmind.ai), licence CC BY 4.0. Link the canonical URL.

Charts and interactive widgets are omitted here; their data is in the JSON above.

---

## What it does

Intelligent Internet is an open-source AI project producing downloadable models, agent tooling and training datasets, plus a hosted II-Agent product with a paid tier, with a planned three-layer blockchain that hasn't been built yet. The distinction matters. What exists today is actually useful. What's promised requires a leap of faith.

Shipped deliverables include II-Agent, an open-source coding agent (over 3,000 GitHub stars, Apache 2.0). II-Medical, a family of clinical reasoning models fine-tuned on Qwen3 (8B and 32B parameter variants, downloadable from HuggingFace in GGUF format for local running). II-Search, a search-optimised 4B model. CommonGround, a multi-agent orchestration protocol. II-Commons, a self-hostable knowledge base. Twenty datasets published on HuggingFace, including II-Medical-Reasoning-SFT (2.2 million samples) and II-Thought-RL-v0 (342,000 reasoning examples). Nine models total. All Apache 2.0 licensed.

On the planned blockchain side: a three-layer architecture. Foundation Layer (L0), a fork of Bitcoin Core v25 with BFT consensus replacing Proof-of-Work. Culture Layer (L1), sovereign roll-ups operated by "National Champions" settling in local Culture Credits. Personal Layer (L2), running on user devices where AI agents execute workflows locally. Dual-currency economy with Foundation Coins (FC, 21 million hard cap mirroring Bitcoin) and Culture Credits (CC, elastic national currencies reserve-backed by FC). Proof-of-Benefit consensus ties every minted coin to verified useful AI work. None of this exists beyond a whitepaper published in July 2025.

Founded by Emad Mostaque through L42 Ltd, a private limited company incorporated in England and Wales in April 2024 (Companies House number 15644961). Mostaque is the sole director. He previously co-founded Stability AI, which released Stable Diffusion (downloaded 100 million-plus times), raised $174 million achieving a $1 billion-plus valuation, and then imploded. He resigned as CEO in March 2024 under investor pressure.

Estimated headcount is 18 people (PitchBook), with core ML development handled by a Vietnam-based team of 6-8 engineers. No funding has been publicly disclosed. The project may be self-funded from Mostaque's personal wealth.

## The product in practice

Grounded in our Intelligent Internet review, its GitHub and HuggingFace repos, and the II community Discord (the build and support channels). There is no token yet, so this assesses the shipped open models and the live hosted product.

**Who it's for.** Two audiences. Developers and researchers who want open, Apache-2.0 models and agent tooling to download and self-host, with a lean toward technical and medical use (II-Medical, II-Agent, II-Search, and twenty published datasets). And, increasingly, ordinary users of the hosted II-Agent product at agent.ii.inc, a web and mobile app with a paid Pro tier, a credit system, and an II Factory mode for generating media.

**What's actually live.** More than the whitepaper framing suggests. Alongside the open models on HuggingFace and GitHub (II-Agent draws a few thousand stars), there is a live, monetised hosted agent: web plus iOS and Android apps, a credit-based Pro subscription, and the II Factory pipeline, with an active team answering support. People build substantive things on it, and in our July 2026 read of the community's show-your-builds channel one non-technical member described assembling a whole local multi-agent system "just figuring it out with ii-agent". None of the token-shaped parts are live.

**Zenith, the newer bet.** Since May 2026 II has also shipped [Zenith](https://github.com/Intelligent-Internet/zenith), an open-source (Apache-2.0) orchestration layer that runs Claude Code, Codex or Hermes as a multi-agent orchestrator for tasks that run for days. It now heads II's homepage under "Intelligence you own". II's 23 September post reports Zenith running DeepSeek V4.1 Flash passing GPT-5.6 Sol on its own AutoResearchExam benchmark at a fraction of the cost. That's II grading its own system, and II itself calls it a system result that can't separate Zenith's contribution from the model's.

**The friction, attributed.** The hosted product's rough edges surfaced in the II Discord support channel in our July 2026 read: subscription credits that do not roll over after a downgrade, occasional mobile login failures, and, most tellingly, credits burned on failed or looping generations, with one user reporting roughly four thousand credits lost when II Factory kept reintroducing fixed issues in a video edit. Useful and shipping, with the credit economics as the recurring complaint.

**The network is a whitepaper.** The three-layer blockchain, the Foundation Coin and Culture Credit economy, and the Proof-of-Benefit consensus that would tie minted coins to verified AI work exist only in a July 2025 whitepaper. None of it is built, there is no token, and the design forks Bitcoin Core with heavy modifications that sit a long way from shipping. The community reflects the gap, with mission-driven builders in one channel and an airdrop-expectant crowd in another, waiting on a token that does not exist yet.

**Where the attention has gone.** II's published output since July is Zenith, a run of Common Wealth essays, and "Champions" (4 September): local public-benefit companies owned through ordinary equity, with the stated aim of listing each on public markets. None of those pages mentions Foundation Coin, Proof-of-Benefit or a blockchain, though the whitepaper describing them is still live at ii.inc. Our read: the chain has slipped further down the queue. Nothing announces its cancellation, and a testnet announcement is our trigger to re-review.

**The vision, and the caveat.** The idea is striking: a currency backed by useful AI work instead of energy or debt, with open models as the public good underneath. The caveat is the founder, since Emad Mostaque built Stable Diffusion and also presided over Stability AI's implosion, and II is thinly funded with a small team, so execution risk on a Bitcoin-scale blockchain is high. Use the models and the agent now; treat the network as a bet on the roadmap and on Mostaque delivering it this time.

## Value proposition

The value proposition splits cleanly into what exists today and what's promised.

What exists is substantive for a pre-token project. According to Intelligent Internet's published model cards, II-Medical-8B achieves 70.49% average across 10 medical benchmarks, performance comparable to models four to nine times its size. The updated version (II-Medical-8B-1706) scores 46.8% on HealthBench, comparable to Google's MedGemma-27B. These models run locally on consumer hardware via llama.cpp, Ollama or LM Studio. Once downloaded, no server dependency, no API key, no one can take them away. For the sovereignty thesis, this is the purest form of "own your mind": a clinical reasoning model running on your own machine.

According to Intelligent Internet's benchmarks, II-Agent scores 61.8% on Terminal Bench 2, above Claude Code at 58.0%. II-Search-4B hits 91.8% on SimpleQA (versus the base Qwen3-4B at 76.8%). These are self-reported results; the models are publicly downloadable and testable. The agent runs. You can verify this yourself.

Training datasets are a concrete public good. Intelligent Internet describes II-Thought-RL-v0 as the first large-scale multi-task RL dataset. II-Medical-Reasoning-SFT's 2.2 million samples enable anyone to train their own medical reasoning models. This contribution to the open-source AI commons is concrete and lasting.

What's promised is the most ambitious vision in the DeAI space. A Bitcoin-inspired blockchain where every minted coin proves useful AI work was performed. Sovereign roll-ups operated by national-level validators. A Living Constitution Assembly with sortition-selected jurors. An Oracle Council of 15 AI agents governing protocol parameters. The whitepaper is technically detailed and intellectually coherent. It's also entirely theoretical.

A counter-narrative exists. Mostaque has form for vision that outruns execution. Stability AI had a spectacular vision too. The pattern of ambitious promises, rapid early progress, and eventual governance and financial challenges is well documented. Whether that pattern repeats is the central question.

## Tokenomics

No token exists. No chain exists. Everything below describes whitepaper design, not operational reality.

Foundation Coins (FC) mirror Bitcoin's supply model: 21 million hard cap, initial block subsidy of 50 FC, halving every 210,000 valid blocks (two epochs). Faster emission than Bitcoin, with roughly 99% circulated within 12 years. The critical difference: blocks are only valid when they contain a Proof-of-Benefit receipt proving useful AI work was performed. Four genesis benefit classes: compute-inference, compute-training, data-curation and agent-orchestration.

No pre-mine is described in the whitepaper. Coins are minted only through validated work. If implemented as described, this would be one of the fairest distribution methods in crypto. The design is structurally closer to Bitcoin mining than to the ICO, VC-backed or insider-heavy models that dominate DeAI. A Treasury receives a programmable share of block rewards (percentage TBD), which is the potential centralisation vector.

Culture Credits (CC) are the second currency: elastic, reserve-backed by FC, and issued by National Champions for local economic activity. Each sovereign roll-up operates its own CC with an on-chain AMM for CC-to-FC swaps. Reserve requirements ensure backing. This dual-currency model is novel and completely untested.

Key parameters remain undefined: exact stake amounts for National Champions, slashing percentages, Treasury burn-vs-grant splits, CC reserve requirements, evaluation thresholds for Proof-of-Benefit. The whitepaper honestly labels these as TBD rather than fabricating numbers, which is at least transparent.

**Warning:** an unofficial Solana token using the ticker "II" exists on CoinSwitch with zero liquidity. This is not endorsed by the project. No official token has been listed on CoinGecko or CoinMarketCap. Do not buy it.

## How to participate

**Use II-Agent.** The most accessible entry point. Available as a web and mobile app (agent.ii.inc) or self-hosted via Docker and CLI from GitHub. Supports multiple LLM backends (Gemini 3, Claude Sonnet 4.5, GPT-5). Self-hosting is free under Apache 2.0 (bring your own LLM API keys); the hosted app adds a paid Pro tier with credits and the II Factory media mode on top of a limited free tier.

**Download and run models locally.** II-Medical-8B in GGUF format runs on any machine with 8GB-plus RAM via llama.cpp or Ollama. II-Search-4B is even lighter. MLX format available for Apple Silicon. This is the highest-sovereignty participation mode: local inference with zero server dependency. Not intended for clinical use.

**Use the datasets.** Train or fine-tune your own models using the 20 published datasets. II-Thought-RL-v0 (342,000 reasoning examples) and II-Medical-Reasoning-SFT (2.2 million samples) are particularly valuable for anyone building AI training pipelines. Requires GPU infrastructure and ML expertise.

**Contribute to open-source repos.** The GitHub organisation lists 24 repositories, none flagged as a fork.  Primary projects: II-Agent (coding agents), II-Researcher (search and research), CommonGround (multi-agent orchestration), II-Commons (knowledge base), II-Thought (RL dataset tooling). Python and TypeScript skills required.

**Validate (planned, not yet available).** The whitepaper describes National Champions: validators requiring economic FC stake (amount TBD), hardware attestation, symmetric 10 Gbit/s bandwidth on three independent physical paths, sub-300ms interlink latency and signed heartbeats every second. Twelve Champions at genesis, long-term target of one or more per sovereign nation. No selection process has been described and no timeline published.

## Honest assessment

### What works

The open-source output is shipped and competitive. Ten models and twenty datasets on HuggingFace (September 2026).  A functional coding agent that Intelligent Internet reports benchmarks above Claude Code on Terminal Bench 2. II-Medical-8B punching above its weight at 8B parameters is technically impressive. All downloadable, all self-hostable, all Apache 2.0. The training methodologies (three-stage medical RL pipeline, four-phase search training with Code-Integrated Reasoning, DAPO safety RL) are documented and reproducible. Third-party GGUF quantisations exist (DevQuasar), confirming community adoption. The whitepaper, whatever you think of its feasibility, is the most technically detailed and intellectually ambitious design document in the DeAI space. The Proof-of-Benefit mechanism tying currency minting to verified useful AI work is a credible conceptual innovation.

### What doesn't work

There's no chain. No token. No validators. No governance. No live testnet. Such public timeline as exists is self-reported: in an April 2026 interview Mostaque floated a Phase 0 spec-freeze and a roughly September 2026 mainnet target. That target has now lapsed without a launch: as of 4 October 2026, II's releases page shows no chain, token or mainnet announcement since August, and no blockchain code exists in its GitHub organisation.

The project is a small company (L42 Ltd, sole director: Mostaque) with an impressive whitepaper and demonstrably useful AI tools. It isn't a protocol. The gap between the vision (sovereign AI infrastructure for every nation) and the current reality (roughly 18 people, no blockchain code, no disclosed funding) is vast.

At our March 2026 review, eight of the 22 GitHub repositories were forks of other projects (OpenAI Codex, Google Gemini CLI, litellm, verl and others), which inflated apparent scope; the organisation now lists 24 repositories and none is flagged as a fork (GitHub API, September 2026).  The language models are fine-tunes of Alibaba's Qwen models, and the newest release, Evoke-Model-Beta-1, is a quantised IBM Granite embedding model, so none is a foundation model trained from scratch. The ML contribution is in training methodology rather than base model capability. If Qwen's licence or availability changes, the model pipeline is affected.

"National Champions" as a concept, one validator per sovereign nation, 12 at genesis with extreme hardware requirements, has no identified participants, no selection process and no operational infrastructure. The Living Constitution Assembly, Guardian Lattice, Oracle Council and progressive decentralisation roadmap exist only as whitepaper text.

### The risk

Emad Mostaque is an extreme concentration of both opportunity and risk. The opportunity: he proved with Stable Diffusion that open-source AI can compete with proprietary alternatives. His 290,000 X followers, AI industry profile and policy engagement (FII9 alongside Peter Diamandis) provide distribution that most DeAI projects cannot match.

His Stability AI track record includes a co-founder fraud lawsuit (Cyrus Hodes, July 2023), a Forbes investigation finding misleading claims about education credentials and institutional partnerships (UNESCO, OECD, WHO and World Bank all denied partnership claims), investor revolt from Lightspeed Venture Partners, and $153 million projected losses against $11 million income in 2023. The question is not whether these things happened. They did. The question is whether the pattern repeats.

No funding has been disclosed for Intelligent Internet. Building a novel L0 blockchain with BFT consensus, 12-plus validators, dual-currency AMMs and a Guardian Lattice requires significant capital. The custom Bitcoin Core fork approach means building everything from scratch: wallets, explorers, bridges, DeFi integrations, developer tooling. Most DeAI projects build on EVM, Cosmos or Solana where ecosystems already exist. This choice is either principled or impractical. Possibly both.

A further complication: the FII Institute connection (Future Investment Initiative, run by Saudi Arabia's Public Investment Fund) adds geopolitical complexity. Sovereign AI infrastructure affiliated with a sovereign wealth fund raises questions about the "sovereignty" framing that haven't been addressed.

### My position

I don't hold any Intelligent Internet tokens because no token exists. I've downloaded and tested II-Medical-8B locally. The models work. The agent runtime works. Whether the blockchain vision materialises is a separate question from whether the open-source AI output is useful, and on the latter, the answer is yes.

## Freedom Score: 42/100

Intelligent Internet scores 42/100 (D grade). Full methodology at [Freedom Score Methodology](https://ownyourmind.ai/resources/freedom-score-methodology/).

**Infrastructure decentralisation (5/20):** No decentralised infrastructure exists. No blockchain, no validators, no consensus mechanism. However, the self-hostable models (II-Medical-8B GGUF, II-Search-4B, II-Thought-1.5B) and self-hostable tools (II-Agent via Docker, II-Commons with PostgreSQL) provide functional infrastructure decentralisation at the model layer. Once downloaded, these run entirely on user hardware with zero dependency on Intelligent Internet's servers. The planned blockchain (Bitcoin Core fork, 12 National Champions, BFT consensus) is ambitious but entirely theoretical.

**Governance decentralisation (2/20):** All decisions made by Mostaque as sole director of L42 Ltd. No governance mechanism exists. No token voting, no DAO, no community decision-making. The whitepaper describes progressive decentralisation (multi-sig to Oracle Council to Living Constitution Assembly) but none of it is implemented. Apache 2.0 licensing provides a form of soft governance: anyone can fork any repo without permission. Meaningful, but it's not governance.

**Token distribution fairness (7/15):** No token launched. The whitepaper design describes Proof-of-Benefit mining with no pre-mine, mirroring Bitcoin. If implemented as described, coins are minted only through verified useful AI work. Structurally fairer than most crypto launches. However, the Treasury receives a programmable share of block rewards (percentage TBD), initial multi-sig control by "founding Champions" means insider control before decentralisation, and key distribution parameters are undefined. Scored generously for design intent with a significant discount for nothing being implemented.

**Censorship resistance (8/15):** The downloadable models provide meaningful censorship resistance today, independent of any blockchain. II-Medical-8B, II-Search-4B and II-Thought-1.5B are complete model weights, not API access. Once downloaded, no entity can revoke, restrict or censor them. Training datasets are published, enabling full reproducibility. II-Agent supports multiple LLM backends and can be self-hosted. However, the web app (agent.ii.inc) is a centralised service. No censorship-resistant network layer exists yet.

**Data sovereignty (7/15):** Self-hosted models run entirely on user hardware with no data leaving the device. II-Agent can be self-hosted via Docker. II-Commons deployable with local PostgreSQL. Training datasets openly published. The planned L2 Personal Layer describes privacy-preserving local enclaves. However, the hosted web app processes data through centralised servers. No zero-knowledge proofs implemented. No confidential computing.

**Open source and transparency (13/15):** The project's strongest dimension. Original repositories under Apache 2.0. Ten models with full weights on HuggingFace (September 2026). Twenty datasets published openly. Training methodologies documented and reproducible. GGUF and MLX formats for local running. Companies House registration public. Deductions: 8 of 22 repos were forks at our March 2026 review (inflating apparent activity; the organisation now lists none), Mostaque's documented history of misleading claims at Stability AI is a transparency concern, and corporate disclosure is minimal (L42 Ltd files micro-company accounts at Companies House with no profit and loss detail).

### Path to improvement

Three changes would materially increase Intelligent Internet's score:

1. **Launch a testnet.** The single biggest signal that the blockchain vision is operational rather than aspirational. A working testnet with Proof-of-Benefit consensus, even with a handful of validators, would demonstrate that the whitepaper design translates to operational code. Until then, the protocol layer is a PDF.
2. **Publish a funded roadmap with milestones.** No mainnet timeline exists. No disclosed funding exists. A transparent roadmap with funded milestones and verifiable deliverable dates would address the most obvious gap between the ambition and the evidence. Construction projects don't get built without a programme. Neither do blockchains.
3. **Identify the genesis National Champions.** Twelve validators at genesis is a specific claim. Who are they? What hardware do they have? Where are they located? Publishing even a shortlist with hardware attestation would convert a whitepaper concept into a concrete commitment.

## Returns: N/A (pre-token)

Intelligent Internet has **no Returns Score**; it is N/A. The Returns Score measures a token you can buy, hold and exit, and Foundation Coin is a whitepaper design with no live chain, no minting and no market, so the dimensions a Returns Score would rest on do not yet exist. Freedom is still scored (above). Full policy at [Returns Score Methodology](https://ownyourmind.ai/resources/returns-score-methodology/) under Pre-Token Projects. What follows is a forward-looking read on the token *design*, not a score.

**Token utility (design):** No token exists. The whitepaper describes Foundation Coins as the medium for Proof-of-Benefit consensus, where every minted coin proves useful AI work was performed. Four genesis benefit classes are defined: compute-inference, compute-training, data-curation, and agent-orchestration. The design is intellectually coherent and more purposeful than most token utility models. But it is a design in a PDF, not a functioning token in a live economy: promising as a concept, and entirely unbuilt.

**Value accrual (design):** The value accrual model is entirely conceptual. FC would accrue value through scarcity (21M cap, halving schedule), utility demand (staking for validation, Culture Credit reserve backing), and the Proof-of-Benefit requirement tying minting to useful work. On paper, this is one of the more thoughtful value accrual designs in the space. In practice, there is no blockchain, no consensus mechanism, no validators, and no working implementation of any of these mechanisms. The gap between whitepaper and reality is as wide as it gets.

**Supply dynamics (design):** The 21 million hard cap mirroring Bitcoin is the strongest part of the design. No pre-mine is described; coins are minted only through validated work. If implemented as designed, this would be structurally fairer than nearly every token launch in crypto. The halving schedule with 99% circulated within 12 years creates predictable scarcity. The Treasury receiving a programmable share of block rewards is the centralisation vector, with the percentage still undefined. The scarcity model is the design's strongest element, but the "if implemented" caveat carries significant weight when no code exists.

**Revenue sustainability:** Nothing on a token basis. No blockchain means no transaction fees, no token means no staking revenue, and no marketplace means no trading fees. The open models and self-hosted agent are free under Apache 2.0, while the hosted II-Agent (agent.ii.inc) earns a paid Pro subscription (amount undisclosed), a demonstrated revenue path that reaches no token holder because no token exists. This is flagged for reassessment once a token launches and the monetisation matures.

**Liquidity and access:** There is no token. No exchange listing, no DEX pool, no OTC market. The unofficial Solana token using the "II" ticker on CoinSwitch has zero liquidity and is not endorsed by the project. Without disclosed funding, the capital backing that typically signals future exchange listings is absent. There is simply no tradeable asset of any kind to assess.

### Path to improvement

Three changes would move Intelligent Internet from N/A to a scored, investable Returns:

1. **Build the blockchain and launch a token.** Every dimension of this score is hamstrung by the same fundamental problem: nothing exists. A live testnet with Proof-of-Benefit consensus and mineable FC tokens would transform every category from theoretical to assessable. Without a chain, there is nothing to score.
2. **Establish a revenue model for the AI tools.** Free, open-source AI models are a public good but not a returns proposition. Introducing optional paid tiers, enterprise licensing, or on-chain inference fees that flow to token holders would create a revenue stream. The current model generates goodwill but zero economic value for potential FC holders.
3. **Secure and disclose funding.** Building a novel L0 blockchain with BFT consensus, sovereign rollups, and a dual-currency AMM requires significant capital. Transparent disclosure of funding sources and runway would signal that the project can survive long enough to deliver on its whitepaper promises.

## Score change log

| Date | Score | Change | Reason |
| - | - | - | - |
| 2026-10-04 | Editorial | N/A | Product-in-practice sweep: dated the Discord anecdotes to our July 2026 read, and moved the snapshot stage from testnet to live, since no testnet exists and the live part is the hosted product. Scores unchanged. |
| 2026-10-04 | Data | N/A | The roughly September 2026 mainnet target from Mostaque's April interview lapsed without a launch; II's releases page shows no chain, token or mainnet news as of 4 October. Scores unchanged. |
| 2026-09-26 | Data | N/A | Added Zenith, II's open-source agent orchestrator and current homepage lead, and noted that II's output since July (Zenith, essays, the Champions equity model) never mentions the chain or token. Scores unchanged. |
| 2026-09-25 | Data | N/A | HuggingFace model count refreshed to 10 (was 9) and dated; "corporate financials not disclosed" qualified, since L42 Ltd files micro-company accounts at Companies House with minimal detail. Scores unchanged. |
| 2026-09-24 | Data | N/A | GitHub stars now over 3,000 (was 3,172); the fork count is dated to the March 2026 review since the org now lists 24 repos, none a fork; II's "first RL dataset" line attributed to II. Scores unchanged. |
| 2026-09-23 | Data | N/A | Integrity records re-cut: 1 claim now quotes Intelligent Internet's own pages verbatim and 0 are our own measurements. Scores unchanged. |
| 2026-07-07 | Returns | 23 → N/A | Pre-token Returns policy (house rule): no tradeable asset now means Returns N/A, not a numeric score. Removed the 23/100 score + breakdown; token-design read kept as prose. Freedom (42/100) unchanged. Methodology updated; also applied to Nous. |
| 2026-07-07 | Data | N/A | Corrected the "no revenue / models are free" framing (review body, frontmatter, research JSON, YT source): II runs a live paid hosted product (agent.ii.inc Pro subscription, credits, II Factory), verified via ii.inc and the II Discord. Revenue Sustainability held at 0 on a token basis, flagged for monthly reassessment. |
| 2026-07-07 | Editorial | N/A | Added a "The product in practice" section (per `specs/product-comparison-framework.md`); the II Discord sweep surfaced a live hosted agent (agent.ii.inc, paid Pro tier, II Factory) alongside the open models, while the Foundation-Coin blockchain stays a July 2025 whitepaper with no token. No score change. |
| 2026-06-26 | Data | N/A | DefiLlama LlamaAI harvest cross-check. Confirmed FC remains pre-token and untradeable on any venue (no CEX/DEX listing), consistent with the review. One update: the harvest surfaced a self-reported timeline from an April-2026 Mostaque interview (Phase 0 spec-freeze, ~September 2026 mainnet target), now noted in the body with the standard skepticism since no blockchain code backs it. Scores unchanged (pre-token: Liquidity & Access and Revenue Sustainability pinned at 0). |
| 2026-03-02 | Both | N/A | Initial publish. Freedom 42/100, Returns 23/100. |