GRT (GRT)Price Live Chart, Prediction, Market Cap, Token Unlocks, GRT Concept Stocks CoinMeta

BTC

$76,918.00

+0.11%

/GRT Price

GRT

No.195

$0.01616

-2.47%

≈$0.02

Market Cap

$185.3M

Cir. Cap

$172.99M

Cir. Supply

10.7B

Cir. Rate

93.3538%

Total Supply

11.47B

Max Supply

10.06B

24h Volume

$858.72M

24h Vol (BTC)

13.86M

24h Turnover

8.012432%

Market Share

0.01%

Performance

Low

0.02

Range

+0.11%

High

0.02

Listing

ATH (2020-12-18)

$2.88

-99.44%

ATL (2021-02-12)

+0%

Official

Contract

Ethereum: 0xc944...da44a7

Official

Whitepaper

github

Social

Network

Converter

OKX Exchange

A leading cryptocurrency trading platform. Sign up to get a 50 USDT crypto mystery box!

Binance Exchange

Binance is a world-leading cryptocurrency exchange platform. Easily buy crypto on your phone.

Chart

Market

About

Huobi Invite: Earn points & commission

Price

Cap

K-line

Depth

1 Minute

5 Minutes

15 Minutes

1 Hour

4 Hours

1 Day

TradingView

+0.94%

24H

-2.47%

+19.79%

30D

+0.69%

-83.34%

All

AI Assistant

GRT下跌原因

GRT资金流向

GRT买卖支撑位

GRT多空比分析

GRT趋势分析

Ask AI

GRT Market

Pairs Funds

Capital Distribution

15m

30m

0 GRT

Outflow

Inflow

0 GRT

Net inflow of funds in 24 hours (GRT)

Net inflow of funds in 24 hours 0

Historical Fund Flow (GRT)

Day

Week

Month

1D Net Inflow

1W Net Inflow

1M Net Inflow

#	Unlock Date	Unlock Amount	Unlock Value	Total Supply %	Unlock Details
1	2026/09/17	12.84M	$207.49K	0.13%	1
2	2026/10/17	12.84M	$207.49K	0.13%	1
3	2026/11/17	12.84M	$207.49K	0.13%	1
4	2026/12/17	12.84M	$207.49K	0.13%	1
5	2027/01/17	12.84M	$207.49K	0.13%	1

About GRT

The Graph (GRT) is a cryptocurrency with market cap $0.3B, ranked #150.

66%

33%

Bullish

Bearish

Community

Yohei

08-03 12:43

BabyAGI underwent nine iterations in three years, but @yoheinakajima consistently focused on graphs. @aiDotEngineer released the final result on YouTube – “Active Graph Agent Runtime (BabyAGI 4)”. This legendary figure, Yohei, is the managing partner of Untapped Capital. This presentation demonstrated the feasibility of building agents around immutable event logs instead of LLMs, accompanied by code, reference agents, and experimental results. - The log is the agent. An immutable, typed event log records the agent's actions and every change that occurs within it, projecting a graph of the agent's state. - Behaviors replace control loops. They react to changes in the graph and emit events. LLMs do not communicate with each other, only with shared state, thus features like replay, rollback, and branching are natively supported. - Policies determine what can be changed. Adding research resources is inexpensive, while editing hints may require manual intervention. New facts can demand that nothing contradicts them. - Views manage context in the form of graph queries, providing a subset of the graph that a behavior should see. - Packages, not skills. Memory, identity, tools, keys, chat: each package bundles an object type and behavior, so you can replace one memory package with another. - Runtime, not a framework. He rebuilt ReAct on this foundation. When a goal is created, an idea is added. When an idea is created, the reason is run. - Logs double as memory. In LongMemEval, facts or entities are not extracted; instead, queries are embedded and messages surrounding the hit results are extracted. When his API key runs out on the 350th issue, the run continues from the 353rd issue instead of restarting. - Self-modification with gating. The mechanism categorizes failures, allowing the agent to edit only the parts that match itself, then requires static analysis, sandboxing, and a measured rerun before accepting the patch. The number of loops was reduced from 8 to 13 to 4 or 5, a small but statistically significant improvement. - It can remember which methods failed. In a Kaggle competition, a deterministic Pokémon Trading Card Agent underwent approximately 80 iterations of tuning. 20 to 30 people passed the tests, and all unsuccessful cases were documented. —Old architecture, new staff. Blackboard and Kafka had decades of development experience, while the LLM agent had only three years.

399

Not Elon Musk

07-30 01:54

We've saved companies like Meta, Unilever, and Johnson & Johnson $260 million by uncovering irregularities in over 19 million invoices. Now, we've raised $75 million to extend our guarantee: if we can't find an invoice for $500,000 in overpayment, we'll pay you $10,000. Schedule a demo: ------------------------ How it works ⬇️ The problem: A Fortune 500 company receives one million invoices annually. Most are only a few hundred dollars. To properly audit an invoice, you need to open the contract, pull up the price list, check the purchase order, and find the bill of lading to prove the shipment details match the carrier's claims. All this work is just to secure $200. So nobody wants to do it. These invoices end up in the hands of human auditors. Auditors only audit a fraction (12%) and then blindly pay the rest (88%). That's where the money goes. At a pharmaceutical company, we discovered an uncontracted surcharge that had been collected for two consecutive years by multiple carriers worldwide. This amounted to a staggering $26 million, yet no one questioned it because each surcharge appeared legitimate. Each surcharge was small, making the cost of increasing the surcharges even higher than paying them. Moreover, there were millions of such surcharges. ------------------------ Solution: Freehand's AI reads all the information: every line of every invoice, comparing it to every contract, price list, transaction data, bill of lading, warehouse records, timesheets, and email correspondence between you and the supplier. It also analyzes all past invoices and transaction data between you and that supplier. All this information is stored in a context graph. When an invoice is received, it knows how much you should pay, what goods were actually shipped, whether services were delivered, the contents of the Service Level Agreement (SLA), and the fees the supplier charged you for the same service in the previous quarter. Then, it will proactively do the following: > Draft a dispute and attach evidence > Send an email to the supplier > If there is no reply, call the supplier > Contact your sourcing team via Slack for more information > Follow up until the invoice is corrected > Approve the correct invoice > Make payment using a different currency and tax structure > Record the correct amount in your general ledger Freehand does all of this just to recover $20, as it processes millions of invoices simultaneously, saving companies 5-10% on expenses, and is 100% SOX compliant and auditable. ------------------------- 🚨 Forward and reply “FREEHAND” to receive an AI skills enhancement guide that has helped over 750 unemployed people find more stable jobs.

321

Yohei

07-30 00:56

🆕 ActiveGraph: Logs as Agents My presentation at the AI Engineer Conference is now online!!! 😆 The presentation was about @activegraphai, an event sourcing graph runtime for building persistent, long-running agents: - Introduction - Code Examples - Experiments - Advantages and Surprises Welcome to watch and share with friends who enjoy exploring cutting-edge technologies. If you have any questions, please feel free to contact me!

427

Ark Invest Tracker

07-28 22:53

Brian Arms: Two Billion-Dollar Startups He Wants to Build On-Chain Today When asked what he would create if he could go back to being 22, Coinbase ($COIN) CEO Brian Arms listed two: - On-chain credit or reputation scoring: Utilizing all the data on the blockchain to build a graph structure, similar to Google ($GOOGL)'s PageRank algorithm, to determine an individual's creditworthiness or a company's credibility. - On-chain advertising network: Monetizing attention, like Meta ($META) and Google in the Web 2.0 era, but integrating it into smart contracts. - DeFi protocols could incorporate revenue-sharing mechanisms into their contracts, so that any wallet or broker that sends customers to them receives a percentage of the revenue, all of which is visible on-chain.

485

Yohei

07-23 05:12

497

kaspaG 𐤊

07-05 04:50

It's been a while since I've talked about my favorite blockchain. Last night I had dinner with the CEO of a major cryptocurrency company and we discussed Kaspa. Here's a summary of our conversation. I just recorded it and imported it into my AI program. We agreed that, from a purely computer science perspective, Kaspa's architecture is a structural masterpiece. It replaces the traditional linear blockchain with a directed acyclic graph managed by GHOSTDAG, cleverly circumventing the core throughput limitation of Bitcoin's Satoshi consensus. Bitcoin, to prevent network errors and maintain decentralization, extends block generation time to ten minutes, thus making a structural compromise; Kaspa, on the other hand, processes blocks in parallel at an astonishing speed without sacrificing security or its pure, fair-start proof-of-work framework. However, in the global macroeconomic arena, technological advantage is rarely the sole determinant of market dominance. History is replete with examples of highly complex protocols ultimately losing to weaker systems with stronger network effects. For Kaspa to win this race, it must overcome enormous psychological and structural hurdles far beyond the code itself. The primary obstacles are the "Lindy effect" and institutional inertia. Bitcoin has withstood nearly two decades of state bans, internal ideological struggles, and relentless crypto censorship. Its main value proposition for global capital lies in its rigidity. Bitcoin's extreme tampering makes it a predictable and stable anchor. Institutional investors buy Bitcoin not because of its speed, but because, like digital gold, it treats low transaction speed as a security feature, leaving speed to other layers. Therefore, any attempt to surpass Bitcoin's "golden" characteristics is doomed to failure. Kaspa's true path to dominance doesn't lie in winning the value store debate from the outset. First, it must completely eliminate Proof-of-Stake (PoS) smart contract chains. With the activation of the Toccata upgrade, Kaspa directly incorporates native programmability, advanced contract logic, and zero-knowledge capabilities into its high-speed Proof-of-Work (PoW) framework. This allows it to directly attack the fatal flaws of networks like Ethereum and Solana, targeting their inherent centralized risks, validator politics, and economic vulnerability under high network loads. If enterprise applications and decentralized finance protocols migrate to Kaspa because of its faster speed, lower cost, and higher security, then developer attention and on-chain liquidity will naturally increase. Bitcoin's current dominance is due to its immutable, decentralized ownership layer, not because its engine is the fastest.

363

MartyParty

06-29 13:46

Everyone's been asking me what @world_xyz is. Now I can finally reveal the answer: World is a proxy-based, modular, intent-centric settlement layer based on the x402 protocol for trustless RWA tokenization. It unifies cross-domain liquidity into a composable state graph and provides institution-grade finality through parallelized zkVM, restaking shared security mechanisms, and MEV internalized solvers, as well as a full-chain, account-abstracted, permissionless verifiable execution environment.

309

Elon Musk

06-19 15:25

Here's a changelog for Grok Build since its release, and everything you need to know: Grok Build is rapidly transitioning from a coding command-line interface (CLI) to a fully-featured terminal-native agent workspace. Since its release, it has added or improved features such as planning/reviewing/approving workflows, cleanup diffs, project-aware context via AGENTS.md, skills, hooks, plugins, an MCP server, parallel sub-agents, headless mode, ACP support, Web/X search, image and video tools, compression, memory management, and long-running sessions. The biggest upgrade users won't want to miss is the rendering layer. Grok Build can now display more technical output directly within the terminal: mathematical formulas, LaTeX, Mermaid diagrams, ER diagrams, UML/class diagrams, state diagrams, sequence diagrams, tables, media output, and richer terminal views. This is crucial for research, machine learning, simulation, algorithms, database design, infrastructure diagrams, paper implementations, and rigorous code review. The terminal is no longer just a place to run commands. It's gradually becoming a place where you understand your work, check logic, view charts, and continuously advance projects without constantly copying output to other applications. The workspace layer has also received a significant upgrade. The agent dashboard lets you manage multiple coding sessions on a single screen, seeing which sessions are running, idle, blocked, or waiting for input, viewing the latest output, replying online, and distributing new tasks as you switch between sessions. The plugin marketplace transforms Grok Build into a scalable developer environment. Plugins can bundle skills, slash commands, agents, hooks, MCP servers, and LSPs. Initial partners include MongoDB, Vercel, Sentry, Chrome DevTools, and Cloudflare, as well as superpowers. Plugin installation can now be resolved directly from registered marketplaces, no longer limited to local paths. The latest version primarily focuses on improving the reliability of all these features in real-world scenarios. Long response times can be recovered after network outages. MCP servers recover better after packet loss or output anomalies. Compression operations no longer hang indefinitely. Notifications only trigger when user attention is truly needed. Linux clipboard support is more robust. Rendering in Windows and iTerm is smoother. Long sessions can be scrolled, resumed, and exited without crashing. Grok Build is becoming a fully-featured terminal-native agent workspace: supporting multiple sessions, plugin-driven architecture, MCP connectivity, graph awareness, mathematical operations, media processing, and long contexts.

407

Wu Blockchain

06-17 23:18

Digital advertising startup EarnOS has raised $6 million in Pre-A funding, led by 1kx, with participation from Coinbase Ventures, Circle Ventures, and Social Graph Ventures. EarnOS stated that its app, ero, has finished its beta phase and will launch in the US, Canada, Australia, and the UK. It aims to help brands validate human internet traffic, reduce bot-induced waste, and reward “genuine digital behavior.” EarnOS also announced a $12.5 million, four-year, non-dilutive strategic investment from Verona (formerly XION), a Layer 1 blockchain that partially supports ero's validation, onboarding, and rewards distribution. (The Block)

460

Yohei

06-06 10:42

This is a deep research agent based on ActiveGraph, which provides complete graph/tracing information such as claims, origins, agent activity, etc. Below is an example of an ActiveGraph-encoded agent. Basically, you always automatically receive tracing information and graphs.

469