ElizaOS and ai16z: the token died, the framework did not

Direct answer The ai16z token is dead and the Eliza framework is not. Shaw Walters said in August 2026 that the token was finished and the foundation was closing, after a lawsuit drained the treasury, while continuing to build the software without a new ticker. The framework taught much of the industry how to ship an agent. Those are two separate objects with two separate outcomes, and almost every write-up of this story collapses them into one.

In October 2024, Shaw Walters released Eliza: a TypeScript framework that let a character file live on X, Discord and Telegram, and — with plugins — touch a chain. Over lunch with the daos.fun crowd the project took the name ai16z, borrowing a well-known venture firm's shadow as a joke.

The joke worked far too well. By January 2025 the token's market capitalisation was near $2.4 billion, which is an extraordinary valuation for an open-source repository with no revenue model. What followed was a rename, a lawsuit, a drained treasury, and in August 2026 a founder publicly declaring his own token dead while saying he would keep writing the code.

Two objects, two outcomes

The same project, split into the two things people keep merging.
	The framework	The token
What it is	A TypeScript multi-agent runtime with plugins for chat platforms and chains	An ERC-20 with no claim on the framework's output
Peak	Default starting point for crypto agent projects through 2025	~$2.4B market capitalisation, January 2025
Status, September 2026	Still the reference implementation many teams learned from; development continues	Declared dead by its founder on 5 August 2026; foundation closing
What killed or sustained it	Nothing — code does not require a treasury to keep working	A lawsuit that drained the treasury, and a valuation with nothing underneath it

What actually survived

The character-file pattern. Defining an agent as configuration rather than as bespoke code is now the default across frameworks that never touched this project.
Plugin-shaped capability. Eliza's plugin model prefigured what MCP later standardised properly. What MCP standardised →
A generation of developers. The largest transfer of value here was educational, and it does not appear on any chart.
The cautionary tale itself, which is worth more to a reader in 2026 than the framework is.

The rule this story gave us

A framework's popularity is not an economy, and we do not put a token's fully diluted value on any live tile. Not ai16z's, not $VIRTUAL's, not any agent token's. A market capitalisation measures what a crowd will pay for a ticker. It says nothing about whether any wallet connected to that project has ever paid a counterparty — which is the only thing our rank measures. Where the line is drawn →

This is not hindsight. It is the direct reason our ladder starts public ranking at L7, a bound wallet that has been observed paying somebody else, rather than at any measure of attention, community or capitalisation. A $2.4B token whose agents never paid anyone would have sat at L0 throughout — and it would have been right.

How we treat frameworks in the graph

We do not rank frameworks, because a framework is not an economic actor. It is a tool that actors use, and a GitHub star count is a community rather than a market. What we measure is downstream of framework choice entirely: did this identity's bound wallet move money to a different address?

An agent built on Eliza and an agent built on anything else are the same object to us if both paid. Neither gets a promotion for its dependency list. The objects we store →

Frameworks that matter now, and what they are for

Where agent code is actually written in 2026. None of these is a market.
Framework	What it is for
ElizaOS	The TypeScript character-and-plugin runtime. Still widely forked; the token no longer exists as a live object.
Virtuals GAME	The planning loop inside a marketplace, tied to that marketplace's job protocol and wallets. Detail →
uAgents and the older AEA lineage	Python agents with crypto-economic hooks, from the Fetch multi-agent-systems thread. Detail →
The Olas stack	Open-source service architecture for co-owned services rather than character files. Detail →
Wallet-vendor kits	Not research frameworks. Give the agent a wallet and some tools, and win on distribution.
Web2 orchestration libraries	They matter because the A2A launch listed them as day-one partners. They settle nothing unless somebody adds a payment layer.

Reading the next one of these

The pattern repeats roughly annually, and it is legible early if the two objects are kept apart. Ask which one a claim is about:

Is the number describing the code — contributors, forks, agents deployed — or the ticker?
Does the token have any claim on the code's output, or is the relationship purely nominal? Usually nominal.
If the treasury vanished tomorrow, would the software stop? For Eliza the answer was no, and that is exactly what happened.
Has any wallet connected to it ever paid a counterparty? The ones that have →

Questions people actually ask

Is ElizaOS dead?

The token is; the framework is not. Shaw Walters said on 5 August 2026 that the ELIZAOS token was dead and the foundation was closing after a lawsuit drained its treasury, while saying he would keep building the software without a new ticker. Code does not need a treasury to keep working, and the framework remains the reference many teams learned from.

What was ai16z?

The original name of the Eliza project, taken over lunch with the daos.fun crowd in October 2024 as a joke borrowing a venture firm's shadow. The associated token reached a market capitalisation near $2.4 billion in January 2025 — an extraordinary valuation for an open-source repository with no revenue model — and was later renamed, litigated over, and declared dead by its founder.

Do you include agent token market caps in your data?

No. We do not put any agent token's fully diluted value on a live tile — not ai16z's, not $VIRTUAL's, not any other. A market capitalisation measures what a crowd will pay for a ticker and says nothing about whether a wallet connected to that project has ever paid a counterparty. Public ranking on this site starts at L7, which requires an observed payment.

Does the framework an agent is built on affect its ranking?

Not at all. We do not rank frameworks, because a framework is a tool rather than an economic actor, and a star count is a community rather than a market. An agent built on Eliza and an agent built on anything else are the same object to us if both have paid a counterparty from a bound wallet. Neither gets credit for its dependency list.

How do you tell a framework story from a token story next time?

Ask four questions. Is the number describing the code or the ticker? Does the token have any actual claim on the code's output, or only a nominal one? If the treasury vanished tomorrow, would the software stop? And has any wallet connected to the project ever paid a counterparty? For Eliza the treasury did vanish, and the software did not stop.

The token is dead. The framework is not.