MMN Editor

President Trump holds a cabinet meeting at Camp David – July 31, 2026
President Trump on Friday told reporters that he’s still talking to Ukraine about providing US weapons like Patriot missile interceptors, but he wants to be “very careful” about giving the advanced technology to another country.
He even suggested that Ukraine “can someday turn on you,” though he said he doesn’t think it “would ever happen.”
Trump previously floated that the US would “give a license” to Ukraine to make Patriot defensive missiles during a bilateral meeting with the Ukrainian President at the NATO Summit in Ankara, Turkey, this month.
“We’re gonna give a license to you to make Patriots. That’s pretty cool, right?” Trump told Zelensky.
“This way he can’t complain that we’re not giving him enough. I said, ‘Make ’em yourself’”
WATCH:

However, as The Gateway Pundit reported, Trump revealed on Thursday that he’s “not sure” about giving them the missiles. “It’s a very extraordinary weapon, and we have to be a little bit careful of who we license to. We don’t really license equipment,” Trump told the Financial Times in an interview.
Then, on Friday, Trump was asked about his comments during a cabinet meeting at Camp David, and he said Ukraine and the US have “not agreed to that,” but talks continue.
“Well, it’s a big step,” Trump said, noting, “we have the greatest weapons in the world. We have the greatest drones in the world. People don’t know that. We have the greatest weaponry in the world.”
“In the case of President Zelenskyy, he’d like to have some Patriots. He’d like to have some tomahawks, which are, you know, lethal, one offensive, one defensive,” the President continued, before highlighting the power of the American weapons.
“Nobody else has that capacity. These weapons are incredible. We have to be very careful about letting somebody build other. You know, it’s very important.”
Trump further said, “You have to be very judicious, but you have to be very careful. Now, we have not agreed to that. We’re talking about it, but it’s a hard thing to give away that kind of technology, and I don’t think this would ever happen, but you know, those people that you give that technology, they can someday turn on you.”
“I gave them the javelins. The javelins are unbelievable,” he added. “We have to be very careful about giving out the secrets to those weapons, and they’re not-you can’t copy them. It’s a little like Nvidia. They have a chip. They say you cannot copy it. The complexity is so great you cannot.”
WATCH:

The post (VIDEO) Trump Says he Wants to Be “Very Careful” About Giving Ukraine License to Build American Patriot Missiles – “We Have Not Agreed to That” appeared first on The Gateway Pundit.

Remembering the legacy of J. J. Hanson.

Trump’s Jones Act Waiver Helped Avert Fuel Shortages Across These Key US Regions

The Department of Homeland Security has issued the broadest Jones Act waiver since at least the 1950s, just weeks after the US-Iran conflict erupted. The measure has allowed foreign-flagged vessels to transport fuel, crude oil, fertilizer, and other essential goods between US ports, bypassing domestic shipping restrictions to mitigate supply chain disruptions across various U.S. regions.

The Jones Act, a century-old law requiring goods transported between U.S. ports to be carried aboard U.S.-Built, US-owned, and US-crewed vessels, was immediately waived in the early days of the U.S.-Iran conflict by the Trump administration.

As of Friday, the waiver has entered its 135th day, suggesting that the administration is continuing to protect domestic supply chains and curb energy prices by temporarily easing maritime protectionist restrictions.

A recent Reuters report stated that the Trump administration was considering another extension of the waiver.

The Cato Institute’s Jones Act Waiver Tracker shows that nearly 50 million barrels have been moved since the waiver took effect on March 17.

The tracker indicates that the waiver sharply expanded domestic fuel movements, allowing foreign-flagged vessels to supply regions where limited U.S. shipping capacity, high transportation costs, and vessel shortages had constrained trade.

After 135 days, waiver shipments to the West Coast totaled 13.9 million barrels, equivalent to 108% of the region’s projected full-year baseline. Puerto Rico received 5.02 million barrels, equal to 134% of its normal annual pace, while New England’s 1.25 million barrels represented an increase of 215%. Gulf Coast shipments were 426% above the projected baseline, albeit from a much smaller historical base.

The data suggest the waiver is unlocking trade routes that were previously uneconomical or unavailable under Jones Act restrictions.

The largest effects are being seen in isolated or import-dependent markets, particularly the West Coast, Puerto Rico and New England, while the impact on the well-supplied Lower Atlantic remains marginal. Cargoes have included gasoline, diesel, jet fuel, renewable diesel, propane and ammonia.

Without the waiver, import-dependent markets such as those in California would likely have faced severe fuel shortages, with gasoline prices potentially surging well above $10 per gallon. The quick action by the Trump administration and the Energy Department appears to have mitigated supply disruptions in those regions where fuel shipments expanded the most.

Recall that suspending the Jones Act was one of six policy options identified by JPMorgan’s head of commodity research, Natasha Kaneva, for how the Trump administration could contain energy costs at the pump. The latest shipment data suggest that this option has played a material role in stabilizing vulnerable regional markets.

Tyler Durden
Fri, 07/31/2026 – 17:20

CBS announced Friday that Tony Romo will be on leave until further notice, one week after he was arrested for operating a vehicle while intoxicated.The network also announced that former Houston Texans defensive lineman J.J. Watt will replace Romo alongside play-by-play announcer Jim Nantz and sideline reporter Tracy Wolfson on CBS’s No. 1 NFL broadcast team.The fact that CBS included Watt in the announcement suggests it expects Romo’s absence to extend into the regular season.CLICK HERE FOR MORE OUTKICK SPORTS COVERAGEThe development has naturally raised questions about Romo’s long-term future with the network. But as OutKick has previously analyzed, those questions existed before his arrest.CBS will televise its next Super Bowl on Feb. 13, 2028, following ESPN’s broadcast this coming season. That game has long felt like the natural deadline for CBS to decide whether Romo remains its lead analyst.For background, Romo signed a 10-year, $180 million contract in 2020. However, CBS is no longer the same company it was when he re-signed. CBS is now part of David Ellison’s Paramount Skydance.Ellison, the son of Oracle co-founder Larry Ellison, has already reshaped CBS News since acquiring the company last year. Most notably, he installed Bari Weiss as editor-in-chief. He has made fewer changes on the sports side, though Paramount recently landed UFC rights in a reported seven-year, $7.7 billion deal.The company paid such a premium for UFC that reports say the NFL wants to negotiate with Ellison before renewing its other television deals, believing he could reset the market with his willingness to overpay.Put simply, Romo’s contract would preclude most executives from demoting him from the Super Bowl with two years left on his current deal. The Ellisons are different. Larry Ellison is currently the world’s sixth-richest person with an estimated net worth of $203 billion.RYAN CLARK MADE IT TOO DIFFICULT FOR ESPN TO JUSTIFY KEEPING HIM AFTER YEARS OF CONTROVERSYAgain, questions about Romo’s future existed before the OWI. Romo has become part of the story for the wrong reasons over the past three seasons. He frequently sounds unprepared, overly goofy and prone to forced banter with Jim Nantz that rarely enhances the broadcast.The odd catchphrases, strained analogies, random noises and awkward jokes are well below the standard set by the other lead NFL analysts. The gap between Romo and Troy Aikman at ESPN, Tom Brady at Fox and Cris Collinsworth at NBC has become increasingly noticeable.It’s difficult to imagine someone as ambitious as Ellison wanting Romo to become a distraction during the first Super Bowl under his wing. And given the amount of coverage Romo generates during the season, he would almost certainly be one.We suspect Ellison would still prefer that Romo show improvement this season and not have to remove the $18-million-a-year analyst from the network’s top broadcast team. In theory, that path still exists. Yet it just became much steeper.It’s unclear when Romo will return, and now his OWI arrest and the viral video are also part of the story.There’s also a real possibility that J.J. Watt impresses both viewers and network executives. So far, Watt has shown promise as a color commentator. Put bluntly, he was better at the job than Romo was last season.If Watt performs well to open the season and Romo returns only to continue struggling, CBS would have a difficult time putting Romo in the analyst chair for Super Bowl LXII next season, a decision that would have to be made before next September.Think of it like a struggling starting quarterback who gets injured. If the backup plays better, history shows the better player often keeps the job, regardless of the guaranteed contract.No matter what CBS says publicly in the meantime, Tony Romo’s future is officially in question.

Biden border vs Trump border (screenshot via Fox)
President Trump on Friday responded to the thousands of African migrants invading the Spanish city of Ceuta, which borders Morocco, with a warning for all Americans. 
Spain’s Interior Ministry estimated that 50,000 people crossed the border since Thursday morning.
Video from the invasion on Thursday shows the thousands of African migrants, all men, pouring across the wide-open border.
WATCH

Another video shows thousands of invaders lined up on the beach and crossing into Spanish territory by simply walking around the fence or swimming, with no border enforcement.
WATCH:

Now they’re defecating in the streets, looting stores, and terrorizing the residents.
“It’s terrible. Remember that picture. That’s going to be us in three years if the wrong side gets in… If the Democrats get in, you will not live a very good life,” President Trump said in a statement to Fox News on Friday.
The Fox and Friends panel went on to compare the border under Biden, where nearly 9 million encounters were reported at the southern border, to Trump’s secure border, where fewer than 200,000 have attempted to cross from February 2025 to now.
The nearly 9 million encounters are likely less than half of the real number under Biden.
WATCH:

As The Gateway Pundit reported, European leaders are panicking over the chaos, with Spain’s Socialist Prime Minister Pedro Sánchez declaring that Spain is “under attack.”
Now, Italian Prime Minister Giorgia Meloni and even leftist French President Emmanuel Macron are taking action to protect their own borders from Spain’s radical open borders policies.
MORE:
NEW: France’s Macron Deploys Troops, Aircraft and Drones to the Border With Spain, Vows To Help Socialist Sánchez With the Migrant Invasion of Ceuta

The post (VIDEO) Trump Responds to Spain’s Migrant Invasion with a Dire Warning appeared first on The Gateway Pundit.

Key Takeaways

Kraken is the best crypto exchange overall due to its low fees, extensive data tools and longstanding reputation. If you’re just starting out with crypto, Coinbase offers greater direction, while Uniswap is your go-to choice for a decentralized exchange.
Crypto exchanges face much stricter regulation in the US than elsewhere because they must comply with the SEC, CFTC, FinCEN and sometimes even state-level regulators.
Centralized crypto exchanges are operated by a single entity and come with stronger rail guards, whereas decentralized crypto exchanges run as peer-to-peer networks, offering greater freedom.
We analyzed 23 crypto exchanges, comparing their fee schedules, trading pairs, security framework and overall standing with auditors and users.

Money is not a client of any investment adviser featured on this page. The information provided on this page is for educational purposes only and is not intended as investment advice. Money does not offer advisory services.
Our top picks for the best crypto exchanges

Kraken: Best Crypto Exchange Overall
Coinbase: Best Crypto Exchange for Beginners
Crypto.com: Best Exchange for Crypto Rewards
Uphold: Best Crypto Exchange for Cross-Asset Trading
Webot: Best Crypto Exchange for Automated Trading
Uniswap: Best Decentralized Crypto Exchange

Pros

Widest selection of crypto and trading pairs of any U.S.-compliant exchange
Near-perfect trust scores across security assessments
Staking on 25 assets, flexible and bonded options

Low fees and niche trading options on Kraken Pro

Cons

Unavailable in Maine and New York
High fees for instant purchases

HIGHLIGHTS

Trading fees
1% for instant buy/sell (ACH), 0.25% – 0.00% / 0.40% – 0.10% on Kraken Pro

Kaiko Exchange Score
77 (AA)

CoinGecko Trust Score
10/10

Why we chose it: Kraken is a perennial favorite of crypto investors, with more than 700 cryptocurrencies and 1500 trading pairs available. Arguably the safest centralized exchange for trading your crypto, it enjoys high marks across all major security assessment platforms. In fact, Kraken has never suffered a large-scale hack since it first went online in 2011 — an accolade that very few others of its size can attest to.
Kraken Pro, the exchange’s advanced trading platform, is especially notable for having some of the lowest spot trading fees in the industry and a highly customizable, responsive dashboard that supports more than 25 trading and market data widgets. It also belongs to a handful of platforms that allow margin and derivatives trading, a highly-restricted practice in the U.S.

Pros

Clean interface, easy onboarding and portfolio tracking
Educational tools that reward users with crypto
FDIC insurance (only for USD cash balances) and crypto crime insurance
Broad fiat on-ramp selection

Cons

35% commission on staking rewards for most assets
High fees on simple buy and sell function

HIGHLIGHTS

Trading fees
~2% for instant buy/sell (ACH), 0.40% – 0.00% / 0.60% – 0.05% on Coinbase Advanced

Kaiko Exchange Score
80 (A)

CoinGecko Trust Score
10/10

Why we chose it: Coinbase does an excellent job of lowering the barrier to entry for new investors by focusing on usability and literacy. Its intuitive interface and quick onboarding process make it easy to verify your identity and get started, while educational incentives like Coinbase Earn help you learn about crypto while earning small amounts, easing you into trading.
Security-wise, the exchange offers six types of two-factor authentication and allows customers to store their funds in a vault with time delayed withdrawals. It also holds FDIC protection (applicable only for USD cash balances) and crime insurance policies to protect custodial accounts on cash balances up to $250,000 and against losses from theft, including cybersecurity breaches — the strongest safety net on our list.

Pros

Up to 5% cashback on everyday purchases
CFTC-regulated crypto options and derivatives
High liquidity and market presence
Zero fees on select tokens

Cons

Many benefits locked behind high CRO staking
Confusing tiered rewards model
Limited advanced trading tools

HIGHLIGHTS

Trading fees
~1%-2% for instant buy/sell on app, 0.25% – 0.08% / 0.50% – 0.18% on regular exchange

Kaiko Exchange Score
82 (AA)

CoinGecko Trust Score
9/10

Why we chose it: Crypto.com features two standout rewards programs, both paid in Cronos (CRO), its native token. Its U.S. Visa Signature credit card earns tiered cashback ranging from 1.5% (Basic) up to 6% (Obsidian). Its prepaid (debit) Visa card earns up to 5% back based on how much CRO you stake — from 0% on the entry-level Midnight Blue card, 2% at Ruby Steel ($5,000), up to 5% on Icy White/Rose Gold ($50,000).
While its higher reward tiers require significant CRO staking, the sheer variety of rewards available — competitive staking yields, referral bonuses, rebates based on trading volume and even reimbursements for streaming services — puts Crypto.com on top when it comes to getting more out of your crypto. This is especially salient when combined with the platform’s high liquidity to ensure smoother transactions and robust security infrastructure.

Pros

Seamless swaps between assets in a single step

Highly transparent business model
Support for hard-to-find altcoins

Staking on more than 20 crypto

Cons

Among the highest trading fees on our list
No advanced trading interface
3.99% debit card deposit fee

HIGHLIGHTS

Trading fees
1.80% – 1.95% for BTC and ETH, 2.85% – 3.80% for altcoins, < 0.25% for most stablecoins Kaiko Exchange Score Not applicable CoinGecko Trust Score Not applicable Why we chose it: Uphold is one of the few multi-asset trading platforms, letting users move directly between crypto, precious metals and even U.S. equities without first converting to cash — what it calls “anything-to-anything” trading. It supports this with a refreshingly simple interface designed for portfolio management rather than active trading. The platform offers a broad range of assets, including more than 200 cryptocurrencies, major metals like gold and silver and more than 40 national currencies. It also features staking on over 20 assets with clearly listed rates and no complicated tier systems, which adds a layer of passive income without much friction. Pros 16 built-in trading bots Highly competitive fees Fully compliant Money Services Business in 48 U.S. states Tight spread fees and smooth bot execution on major pairs Cons Volatile spread fees due to external liquidity sources Unavailable in Alaska, Hawaii and New York Limited fiat on-ramps and off-ramps HIGHLIGHTS Trading fees 0.1% maker / 0.5% taker Kaiko Exchange Score Not applicable CoinGecko Trust Score Not applicable Why we chose it: Formerly Pionex.US, Webot’s main appeal is its suite of 16 built-in AI trading bots which are available for free and require no coding or external integrations. These range from basic ones, like bots that oscillate between different order settings or set up recurring buys, to those with advanced strategies, such as Grid, DCA (Martingale) and Rebalancing bots. The platform sources liquidity from major exchanges like Binance and HTX, which helps maintain tighter spreads and smoother execution for bot-driven trades compared to smaller standalone platforms. Its flat 0.05% maker/taker fee is also quite competitive, especially for high-frequency strategies where costs can quickly compound. Pros Thousands of tokens on more than 11 chains Fully decentralized and non-custodial No sign-ups or Know Your Customer (KYC) Earn by providing liquidity Cons No customer service or fiat on-ramp Inefficient for small trades Costs can vary widely HIGHLIGHTS Trading fees 0.3% liquidity fee Kaiko Exchange Score Not applicable CoinGecko Trust Score Not applicable Why we chose it: Uniswap remains the defining decentralized exchange year-over-year, operating as a fully on-chain protocol where users trade directly from their crypto wallets. This model eliminates counterparty risk and has made Uniswap a cornerstone of DeFi, with billions in total value locked and deployments across dozens of blockchain networks. The fourth and latest version of Uniswap introduces features like customizable liquidity pools, dynamic fees and on-chain limit order capabilities, bringing its functionality closer to centralized exchanges while maintaining its decentralized nature. Availability on Layer 2 networks such as Arbitrum, Optimism and Base also makes trading far cheaper, often reducing fees to just cents per swap. Other crypto exchanges we considered The following trading platforms are all strong or well-known options. However, they didn’t make it to our top list because they lack certain core features, are not dedicated crypto exchanges or don’t excel in any particular area. Binance.US Binance.US is one of the most affordable crypto exchanges in the U.S., with trading fees starting around 0.1% and discounts available on certain pairs making it attractive for cost-conscious users. However, regulatory pressure has significantly impacted its operations in the past — USD deposits and withdrawals have been restricted at times, and its product offerings have narrowed — leaving it less competitive than before despite its low fees. Bitstamp Bitstamp, owned by the popular trading platform Robinhood, is one of the longest-running crypto exchanges, having built its reputation around regulatory compliance, security and reliability. It holds licenses across the U.S. and Europe and keeps roughly 95% of customer assets in cold storage with 1:1 backing. Despite its transparency and low fees, a smaller selection of assets and limited advanced features left it trailing other competitors on our list. Gemini Gemini features an intuitive user experience, solid custody practices and support for a moderate range of nearly 100 cryptocurrencies. It also offers an advanced interface for active traders and institutional-grade infrastructure, appealing to more serious investors. The platform’s main drawback remains cost, as its standard trading fees and spreads are higher than many competitors. While the exchange is reliable, it doesn’t stand out strongly enough on pricing or features to lead the market — though this could change now that Gemini won approval from the U.S. Commodity Futures Trading Commission to operate its own regulated derivatives clearinghouse. What you need to know about crypto exchanges Crypto exchanges are the primary gateway for trading digital currencies like bitcoin (BTC) and ether (ETH) for other cryptocurrencies or fiat currencies such as U.S. dollars and euros. They are online platforms that act as intermediaries, allowing users to buy, sell or swap crypto. In addition to trading, crypto exchanges provide a variety of services unavailable elsewhere. Crypto staking is a big one: it allows you to earn crypto back for setting aside a portion of the coins in their wallet. Some even offer cashback card programs and tools for web3 developers. Exchanges are essential to the crypto ecosystem because many traditional investment firms don’t support direct crypto trading. Without them, you would have to find another person willing to trade crypto, agree on an exchange rate and send the coins directly to each other’s crypto wallets — a somewhat convoluted process. The different types of crypto exchanges There are two main types of crypto exchange: centralized and decentralized. Platforms that combine features of both types are sometimes considered “hybrid” exchanges. Each type of exchange is vulnerable to different cyberthreats: Centralized exchanges (CEX) Centralized exchanges (CEX), such as Coinbase, Kraken and Gemini, are the most widely used type of crypto exchange. They are operated by private companies that provide a platform for you to buy, sell and trade cryptocurrencies. CEXs tend to have higher trading volumes, deeper liquidity and a wider range of trading pairs than their decentralized counterparts. Centralized exchanges are often considered a good option for beginners due to their simpler onboarding access to customer support. However, they require you to trust the platform with the custody of your funds and to provide personal information to verify your identity, which some may feel uncomfortable sharing. Decentralized exchanges (DEX) Decentralized exchanges (DEX) like Uniswap and Bisq follow in the spirit of bitcoin by operating with no central point of control. Instead of relying on company-run servers, most modern DEXs run on blockchains, which are maintained by decentralized networks of computers, servers and devices. If centralized exchanges operate as brokerages, decentralized exchanges are more like marketplaces. They facilitate peer-to-peer trading, giving users full control over their assets. This eliminates counterparty risk, but also introduces some drawbacks — namely a steeper learning curve and a lack of traditional customer support. Understanding crypto exchange fees Like exchanging fiat currencies, trades on a crypto exchange usually incur fees, which is how these platforms make the most of their revenue. But it’s not always clear what exactly you’ll be charged for — even less so when exchanges can run on entirely different fee models. Crypto exchange fees can be broadly classified into four categories: trading fees, spread fees, network fees and deposit and withdrawal fees. Trading fees Charged when buying or selling crypto, trading fees are most often seen in spot exchanges under a maker-taker model, where one of two percentages applies to your transactions under different circumstances (for example, 0.20% / 0.40%). The maker fee applies when you add liquidity to the exchange by placing an order that doesn’t immediately fill. The taker fee applies when you remove liquidity, which happens when you place an order that fills instantly at the current price. Many exchanges lower these fees for investors — sometimes eliminating them entirely —based on their trading volume and how much of a certain coin they are holding. Spread fees Some crypto exchanges charge you for the difference between the price at which they sell a cryptocurrency and the price at which it buys it from you. This difference, called a spread, is not always an explicit “fee.” Often it’s embedded in the price of your trade. For example, if bitcoin’s market price is $90,000, an exchange might sell it to you for $90,300 and buy it from you for $89,700. That $600 difference is the spread — effectively a hidden fee you pay on each trade. Spreads are common when using the “instant buy” feature of crypto exchanges. Although convenient, it’s usually more expensive than regular spot trades under a maker-taker model. Deposit and withdrawal fees Deposit and withdrawal fees may apply when transferring money into or out of your crypto exchange, for example, via debit card, credit card or PayPal. These fees vary widely from exchange to exchange depending on the payment method and type of asset you’re moving. Bank transfers are often free or low cost. Credit and debit card deposits can carry fees of 2% to 5%. Crypto withdrawals often come with blockchain network fees (like gas fees on ethereum) and sometimes incur additional platform fees. Network fees Also known as gas fees, network fees are paid to cryptocurrency miners or validators, who process and secure crypto transactions on the blockchain. Most importantly, most exchanges do not set these fees: They’re built into the blockchain itself and are based on network activity. Many centralized crypto exchanges incur this cost, though some may allow you to pay more in fees to speed up your transactions. Users on decentralized exchanges have to pay this fee themselves, since there is no third party between them and the crypto miners. Latest news Bitcoin and other cryptocurrencies continue to struggle despite the Trump administration’s efforts to support the industry. Investors have been moving away from the asset amid persistent inflation, uncertainty over interest rates and broader demand for safer, more established investments. The decline is another reminder that even regulated exchanges cannot protect users from market volatility, and that investors should not assume using one makes crypto itself less risky. BlackRock, Google, Coinbase and several major banks are backing Open USD, a new dollar-backed stablecoin expected to launch later this year on networks including Base and Solana. The consortium reportedly has support from about 140 companies, including Visa, Stripe and Mastercard, signaling that stablecoins are moving beyond crypto trading and deeper into mainstream payments and settlement. Crypto Exchanges FAQs Are crypto exchanges safe to use? Crypto exchanges are generally safe to use, partly because they must follow more stringent regulation today. To operate in the U.S., exchanges have to register with FinCEN and follow anti-money laundering (AML) and know-your-customer (KYC) rules. This doesn't mean they are impervious to bad actors or cyberattacks, but they are held up to a higher standard overall. Can I store my crypto on an exchange? You can store your crypto on an exchange, but keeping large amounts of it there is risky. If the exchange's platform is hacked and your crypto is moved out of its wallet, there is no way to recover it. The same goes for any exchange with no financial assistance for existing clients that shuts down. What payment methods do crypto exchanges accept? Major exchanges accept a range of payment options, including bank transfers (ACH, wire, SEPA), credit and debit cards, third-party services like PayPal and Apple Pay, and other cryptocurrencies for crypto-to-crypto trading?. Some exchanges also support gift cards or transfers from e-wallets. Do I need to verify my identity to use a crypto exchange? Most regulated exchanges require users to complete an identity verification process called Know Your Customer (KYC) before allowing fiat deposits, withdrawals or large trades. You may need to submit your ID, a photo and proof of address as part of this process. Decentralized exchanges often skip this process altogether, extending greater privacy to its users. What happens if I send crypto to the wrong address? If you send crypto to the wrong address, the transaction is typically irreversible and the funds are likely lost permanently. Recovery might be possible if the address belongs to another user or exchange, but that depends entirely on their cooperation. Your funds may also be lost in the event that the address is invalid or incompatible, though some exchanges or wallets can occasionally help recover them in specific cases. Methodology Our team at Money researched over 23 crypto exchanges and evaluated them based on cost, security and user experience. We mainly focused on dedicated crypto exchanges and then looked at trading platforms that also let you trade other securities. Cost Fees can add up quickly, especially for frequent traders. Many platforms offer tiered fee schedules where trading costs drop as your volume increases. Some also provide discounts for using native tokens like BNB or CRO, or even waive fees altogether for high-volume orders. We favored exchanges that clearly establish their fee and explain them in layman’s terms. Security Secure crypto exchanges protect both your funds and personal information through strong preventative measures and responsible account management. The best platforms conduct regular penetration tests, host bug bounties and request third-party audits to identify and fix vulnerabilities in their code. They may also offer insurance to cover losses from internal errors. On the user side, exchanges should enforce best practices like TLS/SSL encryption and multi-factor authentication (MFA) to ensure safe account access. These tools — along with a strong track record of guarding against breaches — are key signs of a trustworthy platform. User experience A smooth experience is essential for beginners and seasoned traders alike. We evaluated platforms based on the design and usability of their mobile and desktop interfaces, favoring those with clean layouts and simple navigation. Strong user experience includes access to educational content like tutorials, guides and explainers to help users understand crypto basics and platform features. We also favored exchanges with research tools like those that provide real-time market data, coin filtering and performance metrics (like market cap and price trends). Summary of our top picks Kraken: Best Crypto Exchange Overall Coinbase: Best Crypto Exchange for Beginners Crypto.com: Best Exchange for Crypto Rewards Uphold: Best Crypto Exchange for Cross-Asset Trading Webot: Best Crypto Exchange for Automated Trading Uniswap: Best Decentralized Crypto Exchange

If you ask an AI coding agent to write a standalone Python script to parse a single JSON file, it will likely give you a perfect answer in seconds. But the same agent often breaks if you ask it to build a systematic data processing pipeline, like ingesting thousands of messy documents, chunking text, scoring quality, and filtering noise for a Retrieval-Augmented Generation (RAG) system that fits your specific enterprise stack.While large language models (LLMs) excel at one-off code generation, their outputs for complex data-processing tasks are typically free-form, disposable scripts. These scripts are detached from the governable workflow abstractions that MLOps teams rely on for production, making them difficult to audit or edit visually.To address this, researchers at Peking University, Zhongguancun Academy, and Shanghai’s Institute for Advanced Algorithms Research introduced DataFlow-Harness, an open-source framework that guides an LLM agent to build structured, visual data-processing workflows step-by-step, rather than writing raw code from scratch.The framework makes AI-generated pipelines easier to manage and integrate into existing architectures because the generated artifacts are persistent and easily editable.The researchers report that the platform achieves a 93.3% observed end-to-end pass rate on a 12-task data-engineering benchmark. Compared to standard Claude Code, it reduces API costs by up to 72.5% and response latency by 49.9%, while achieving nearly the same success rate as an AI given the entire codebase to write standard scripts. For enterprise teams, this means getting the speed of AI automation without accumulating unmanageable technical debt, ensuring that pipelines remain secure, auditable, and ready for production.The “NL2Pipeline gap”Data-centric AI requires workflows for tasks like synthetic data generation, retrieval augmentation, and model training. While LLMs can translate natural language into executable implementations to perform these tasks, high task accuracy is insufficient for production deployment.”The first wall is usually not writing Python,” Runming He, first author of the DataFlow-Harness paper, told VentureBeat. “Modern coding agents can often produce a plausible script quickly. The harder problem is grounding that script in a live production platform: using operators that are actually installed, matching the real dataset schema, referring to registered datasets and model services, preserving dependencies between stages, and leaving behind an artifact that another engineer can understand and revise.”General-purpose AI agents frequently hallucinate dependencies, relying on unavailable operators or outdated platform assumptions. Instead of leaving behind an artifact that another engineer can understand and revise, they generate disposable code that is difficult to audit through workflow managing tools.The researchers define this challenge as the “NL2Pipeline gap”: the disconnect between a user expressing workflow requirements in natural language and the production environment requiring structured and persistent pipeline assets.The researchers demonstrated this gap in their experiments. For example, when Claude Code was allowed to write standard, free-form scripts using codebase context, it hit a 94.2% success rate. However, when restricted to only using the platform’s specific building blocks to create a native workflow graph, its success rate dropped to 83.3%. This gap is the paper’s central finding: native, governable pipelines are meaningfully harder for the agent to produce than throwaway code.“Closing this gap requires more than improving code-generation accuracy: construction must remain grounded in platform semantics and produce artifacts that integrate with the host platform,” the researchers write.How the four components work together”DataFlow-Harness changes the agent’s action space,” He said. “Instead of asking the agent to emit arbitrary code, it retrieves the live operator registry and current pipeline state through MCP and applies typed, incremental changes to a persistent DAG.”To achieve this, the platform organizes workflow synthesis around four components: the Data Pipeline Backend, the interaction layer (DataFlow-WebUI), the MCP Tools Layer, and the AI guidance layer (DataFlow-Skills).The Data Pipeline Backend acts as the authoritative source of truth across conversational, visual, and programmatic interfaces. It represents the pipeline as a directed acyclic graph (DAG), a structured workflow map containing data sources, configured pre-built processing modules (which the researchers refer to as “operators”), and execution dependencies. Instead of generating free-form code, agents interact with this backend through “typed mutations,” like adding an operator or connecting edges.DataFlow-Skills are markdown files that inject domain-specific knowledge into the model’s context window, guiding it on operator-selection patterns, schema inference, and assembly procedures. Rather than letting the AI guess how to assemble components, skills provide the AI with compatibility rules, teaching it how to correctly match different data formats and handle complex data structures without breaking the pipeline. The MCP Tools Layer gives the AI access to the operator registry and current state of the data workflow. The AI proposes structured changes through the tools layer. The system validates the changes to ensure the workflow runs in a valid sequence and that every connected module speaks the same data language.DataFlow-WebUI provides two interfaces that allow humans and AI to build the workflow together. Developers can describe workflow requirements in natural language through a conversational interface. They can also access the workflow as a graphical map in a visual DAG editor. Here, they can directly inspect the changes proposed by the AI and make modifications.“The current implementation performs static checks against platform metadata before accepting pipeline changes,” He said. “These include checks for registered datasets, operators and model-serving references, field flow, and some invalid parameter usage, as well as structural validity. The result is visible in a graphical editor and can be revised either manually or by the agent in later turns.”The results: 93.3% pass rate, 72.5% lower costThe researchers tested DataFlow-Harness on a benchmark of 12 tasks across six industrial data-processing scenarios, such as QA generation, review governance, and schema normalization. They used Claude Opus 4.7 as the backbone model in their experiments.They compared DataFlow-Harness against three baselines:Vanilla CC: An unconstrained coding baseline using standard Claude Code.Context-Aware CC: An agent that has access to the DataFlow codebase in its context window.MCP-only: An agent that has access to the DataFlow MCP tools and is instructed to generate platform-native DAGs (without access to DataFlow-Skills).DataFlow-Harness achieved a 93.3% end-to-end pass rate, improving by 10.0 percentage points over MCP-only and beating Vanilla CC (91.7%), while being within 0.9 percentage points of Context-Aware CC (94.2%).Importantly, it reduced API costs to $0.261 per task, a 72.5% drop compared to Vanilla CC and 42.8% compared to Context-Aware CC. In generating workflows, it was 49.9% faster than Vanilla CC and 17.6% faster than Context-Aware CC.DataFlow-Harness proved particularly effective on complex tasks that depend on implicit domain knowledge, like QA generation. The baseline MCP-only approach frequently generated structurally valid DAGs but struggled to infer task-specific procedures from operator descriptions alone.To show how this works in the real world, the researchers detailed a textbook-to-VQA extraction task. This job required the AI to stitch together capabilities such as PDF parsing, layout recovery, OCR, figure extraction, multimodal understanding, and long-range question-answer matching. DataFlow-Harness achieved 97.2% precision and an 87.3% coverage rate, easily beating the baselines. By having the AI snap together existing platform assets rather than coding complex tasks from scratch, it recovered more valid QA pairs from the document.Their experiments also showed that DataFlow-Harness is highly effective at creating data generation pipelines. For example, in a synthetic instruction-data generation task, the agent built a multi-stage pipeline that generated candidate instruction–response pairs, critiqued and rewrote them, scored them with an LLM-based judge, and filtered low-quality outputs before training.”Such workflows are costly to build and fragile to maintain as collections of ad hoc scripts,” He said. “The harness does not make them automatically safe, but it turns them into explicit, editable stages that engineers can inspect, test, and govern using normal production controls.”Similarly, when tasked with building a math data cleaning-and-synthesis pipeline, the data produced by the DataFlow-Harness pipeline trained a better-performing model with higher average accuracy on AIME24 and AIME25 benchmarks than the data produced by the vanilla Claude Code pipeline.Tech stack fit and implementation tradeoffsFor engineering teams evaluating DataFlow-Harness, it is important to understand how it fits into existing infrastructure. Released under the Apache 2.0 license, the current implementation requires a bit of engineering to fit into popular tech stacks.”The current implementation is native to the DataFlow platform; it is not a turnkey Airflow, Prefect, or Spark plug-in,” He said. To use those systems as an execution backbone, teams must build an adapter to connect their organization’s registry, metadata, and execution interfaces to the agent’s control layer.Furthermore, organizations must invest in the boundaries they want the AI to respect. This requires maintaining an operator registry, defining schemas, and encoding recurring domain procedures as Skills. Because of this overhead, He recommends against using the framework for small, one-off transformations where a simple script suffices, or in legacy environments that cannot expose reliable metadata.Finally, while the platform prevents illogical connections by validating structural properties, it is an engineering control layer, not a compliance substitute. “The harness should still be treated as an engineering control layer, not as a substitute for compliance policy, validated detection models, access controls, audit logging, or human approval,” He said.The platform is open-source, and developers can access the source code and codebase documentation directly via the project’s GitHub repository.As protocols like MCP become standardized, the boundary between human engineers and AI agents will shift. “The goal is not autonomous data engineering without oversight,” He said. “It is a better division of labor: agents perform repetitive construction inside explicit boundaries, while engineers remain responsible for the semantics, policies, and consequential decisions that require domain accountability.”

Alibaba has an agreement for Moonshot to use its Nvidia H200s, according to Bloomberg — but the Chinese tech giant pushed back against the report.

The sandwich chain’s IPO was more than 10 times oversubscribed.