There were two announcements out of the American government this week, and if you read them side by side you would not believe they described the same region.

The first, delivered Thursday evening by President Trump, was a peace. Hamas, communicating through mediators, had agreed to the final text of a plan to decommission its weapons in Gaza, negotiated through an international body the administration calls the Board of Peace. "The threat that emerged from Gaza on October 7 will NOT be allowed to rebuild," the president wrote. Israeli forces would withdraw in phases as disarmament proceeded; an International Stabilization Force and a new Palestinian police force would take over security. It was, officials said, the first time in the war that Hamas had ever agreed to a specific mechanism for handing over its arms.

The second announcement arrived a few hours later, before the sun was fully up over the Persian Gulf. Iran's army said it had launched explosive drones at the Ahmad al-Jaber Air Base in southern Kuwait — a key hub for American aerial surveillance in the region — targeting aircraft hangars, satellite communications, and equipment warehouses. Kuwait's defense ministry confirmed it had detected the drones "since dawn" and destroyed them, with falling debris causing material damage but no casualties. Iran said the strike answered an American attack on a residential home on Qeshm Island, where local media reported three members of one family were killed.

So one government declared an end to one war and prosecuted another in the same news cycle. This is the shape of American policy in the summer of 2026: peace as a document, war as a fact, and the distance between them measured in the hours it takes a drone to cross a border.

Read the Gaza deal carefully and the seams show. U.S. officials could not offer a timeline for Hamas's disarmament. The Gaza police force is meant to turn over weapons within two weeks; the surrender of heavy arms and the decommissioning of tunnels comes "later," in a process officials said could run 200 to 350 days. Movement between phases happens only after a verification committee confirms the previous one is complete. In other words: a framework, refined over weeks, conditional at every step, with the hard parts pushed past the horizon. The Board of Peace calls it historic. It is also, in its own officials' words, "a conditions-based deal" — which is what you call an agreement when you are not yet sure it is one.

Meanwhile the actual war, the one nobody is negotiating, keeps widening. The U.S.-Iran ceasefire collapsed on July 8. Since then the fighting has spread to Kuwait, Iraq, Jordan, Egypt, the Red Sea, and the Caspian. On Thursday the Senate voted, for the thirteenth time, on whether to rein the president in. It failed, 49 to 50. Iran's Revolutionary Guard said Friday it had struck two tankers trying to cross the Strait of Hormuz under American air escort and turned four others around. The strait — which used to carry a fifth of the world's oil — has been effectively closed since February.

The temptation, reading all this, is to pick the story you want and ignore the other. The White House would like you to read the peace. Tehran would like you to read the war. Our position is that you cannot understand either one without holding both in your hand at once: an administration that can, on the same day, broker a disarmament in one theater and escalate in another, and does not appear to see a contradiction.

That is not hypocrisy, exactly. It is the logic of a foreign policy that has decided force and diplomacy are simply two settings on the same machine, to be applied wherever each seems likely to work this hour. Gaza gets the document. Iran gets the drones. The people underneath both — the family on Qeshm, the technocrats being handed a demilitarized Gaza they did not ask to govern, the sailors on tankers turned back at sea — get whichever one reaches them first.

We lead with the gap because the gap is the story. Everything else this issue — the oil that is 65 percent choked, the Fed that cannot agree what the war is doing to prices, the migrants who crossed into Ceuta because Europe's south is one more front — flows out of it.

Strip the announcement of its adjectives and here is what Hamas agreed to on Thursday.

Through mediators, the group communicated that it accepted the final text of a plan that includes decommissioning its weapons — the first time in the war it has committed to a specific mechanism for doing so. That much is real and it is not nothing. Nine months of deadlock over the second phase of the ceasefire had turned largely on exactly this question.

Now the qualifiers. U.S. and Board of Peace officials, briefing reporters, could not give a timeline for the disarmament. What they described was a sequence: within two weeks, the Gaza police force turns its weapons over to the technocratic administration the Board backs. The harder pieces — heavy weapons, the tunnel network, the infrastructure — come later, in a process officials estimated could take between 200 and 350 days. Israeli withdrawal is phased against that progress, on a timetable still to be finalized. And you do not move from one phase to the next until an implementation verification committee certifies the last one is done.

Detailed conditions, officials said, would be published within two weeks. Which means the thing announced Thursday as an agreement is, more precisely, an agreement to publish an agreement.

The skepticism in the region is not hard to find. An Israeli political source told CBS News that Gaza was not even discussed at a recent high-level meeting, and characterized the withdrawal talk as a "trick." Palestinian Islamic Jihad called reports of a factional deal "inaccurate." A former Israeli army chief warned that any deal failing to dismantle Hamas would be a "complete failure." Israel's precondition all along has been full demilitarization *before* any process — not phased against it.

Europe, for its part, welcomed the thing. France and the EU both called it a step toward peace and urged "effective implementation" — diplomatic language for: we will believe it when the tunnels are gone.

The honest read is that a real concession happened and a real agreement did not, at least not yet. Hamas said a word it had never said. Everything that would make the word binding — the dates, the sequence, the enforcement — remains to be written. We will cover it when it is.

The number is almost impossible to hold. In roughly twenty-four hours, about 60,000 migrants crossed from Morocco into Ceuta, the Spanish enclave on the North African coast — a figure equal to some 70 percent of the city's population of about 83,000. Ceuta grew by more than half in a day.

They came by land and by sea, many swimming roughly five kilometers from the Moroccan town of Fnideq, others around a breakwater barrier that juts into the water near the Tarajal beach checkpoint. At least 34 people died, according to the enclave's president, Juan Jesus Vivas — some by drowning, some crushed in a stampede at the breakwater. Some figures ran higher; the interior ministry initially confirmed 19 bodies recovered from the sea. The dead were, overwhelmingly, young men and adolescents.

By Friday, Prime Minister Pedro Sánchez had flown to Ceuta and called the crossings a "violation" of Spanish territorial sovereignty. Spain deployed its armed forces alongside police, reinforced the fence with Morocco, and by afternoon said around 37,500 people had already returned — some voluntarily, many led back through the same breakwater by soldiers. Spain and Morocco appeared to have largely halted the mass crossings.

Ceuta authorities linked the surge to a Spanish Supreme Court ruling earlier this month barring the immediate return of migrants who arrive by sea without due process. The ruling does not cover those who climb the land fence — which is part of why the numbers, and the legal chaos, are what they are.

The shock did not stay in Spain. France announced increased checks on its Spanish border. Italy threatened to suspend Spain from the EU's internal open-border scheme. Germany moved to prolong its own border controls. Ceuta and Melilla are the European Union's only land borders with Africa, and for one day this week the entire architecture of the continent's migration policy was tested at a single spit of land — and visibly bent.

The scene, observers kept noting, recalled 2021, when about 10,000 crossed. This was six times that. What made 60,000 people decide, in the same window, to move is the question Madrid and Rabat will now spend months arguing over. What is not in question is that the seam held only after the army arrived.

On Thursday the Senate did the thing it has now done thirteen times: it voted on whether to make President Trump stop the war with Iran, and it declined.

The vehicle was a motion to discharge Senator Kirsten Gillibrand's war powers resolution from the Foreign Relations Committee, which would have forced the withdrawal of U.S. forces from hostilities with Iran absent a congressional declaration of war. It failed 49 to 50 — one vote short of even reaching the floor for debate.

The arithmetic is worth reading closely, because it is the same arithmetic that has held the war open since the ceasefire collapsed on July 8. Three Republicans — Susan Collins of Maine, Lisa Murkowski of Alaska, and Rand Paul of Kentucky — crossed over to support the measure. One Democrat, John Fetterman of Pennsylvania, crossed the other way to oppose it, as he has done on more than a dozen such votes. Mitch McConnell missed the vote, citing health. The lines have barely moved in weeks of voting.

"This reckless war is driving up monthly bills and making everyday life harder for working families," Gillibrand said before the vote, framing it, pointedly, as an economic argument as much as a constitutional one. After it failed she accused Republicans of choosing a "costly war" over the American people.

Here is the part that gets lost. Even had the motion passed, it was procedural — it would have needed another vote before final passage, and the whole exercise would have amounted to a symbolic rebuke, not a binding leash. The War Powers Act of 1973 has always been more assertion than enforcement. Since the ceasefire broke, four American service members have died from Iran-launched strikes; the president attended their dignified transfer last week. The Senate keeps voting. The war keeps going. Thirteen times now, the second fact has outweighed the first.

The Federal Reserve held its benchmark rate at 3.50 to 3.75 percent this week, and the interesting thing was not the hold. It was that three of the twelve voting members refused to sign onto it.

Cleveland's Beth Hammack, Minneapolis's Neel Kashkari, and Dallas's Lorie Logan all dissented, each preferring a quarter-point hike now. On Friday they made the case in public. "The longer that high inflation persists, the more challenging and costly it can be to bring it back down," Hammack said, adding that she is "not confident" inflation returns to target on its own and that she sees pressure coming from the demand side, not just energy. Kashkari argued that small hikes now prevent larger moves later. This is a 9-3 committee, and the three are not backbenchers.

Chair Kevin Warsh, who sided with the majority, has done something Powell never did: he has thrown away forward guidance entirely. No dot-plot theater, no signaling — just data dependency and a hard line in rhetoric. "There is no soft inflation target," he told reporters. "There's only a target, and it's 2 percent." Which prompted the obvious question, put to Reuters by one analyst: if he is as tough as he says, why hasn't he acted?

Markets are asking the same thing, and they are not reassured. They called it a "hawkish hold" and sold. The Dow fell about 1,153 points, roughly 2.2 percent; the S&P 500 dropped about 1.5 percent. Fed funds futures whipsawed — a 77 percent chance of a September hike, then back to around 57 percent — because when a chair abandons guidance, every meeting becomes a coin flip and every data point a crisis.

And the data itself is a trap. June PCE cooled to 3.7 percent year over year, down from 4.1, with core at 3.3 — a genuine improvement. But it cooled *because* a brief ceasefire pushed energy prices down. That ceasefire is gone. Hostilities have resumed, oil is climbing again, and the very thing that made June look good is now reloading. Warsh himself conceded the point in unusually candid terms, saying monetary policy "does have an important role to play" in addressing "a series of successive supply shocks" that might entrench higher inflation.

So here is the Fed in one sentence: a chair talking like a hawk and voting like a dove, a dissenting bloc that thinks the war makes their case for them, and a June inflation number that was a gift from a ceasefire the war has already taken back. The 10-year yield climbed nearly 27 basis points in July, its biggest monthly jump since March. The bond market, at least, has made up its mind about which way this is going.

Oil rose Friday, and the reason is a good lesson in how a war prices a commodity. Iran's Revolutionary Guard said it had struck two tankers trying to cross the Strait of Hormuz under U.S. air escort and turned four others around. Brent gained about 1.5 percent to roughly $90.34; WTI rose about 2.2 percent past $85. For July as a whole, Brent is up nearly 24 percent, WTI about 23.

What is actually flowing is murkier than the price suggests. Commonwealth Bank of Australia estimates ship traffic through Hormuz has recovered to roughly 30 to 35 percent of pre-war levels — and warns that a rebound to 50 to 60 percent could be enough to flip the market back to oversupply and crack prices. But ships are not the whole story. The U.S. energy secretary says about 13 million barrels a day is leaving the Persian Gulf, with roughly half moving through the strait and half bypassing it via pipeline — which would put total regional flow nearer 65 percent of normal. Much of the strait shuttling happens with transponders off, invisible to tracking.

That gap between what you can see and what is actually moving is why the market cannot settle. Traders are pricing rumor. "Iranian media outlets claim some tankers were forced to turn around," UBS's Giovanni Staunovo noted, "so there are still low flows." On Friday two very large crude carriers did exit the strait, per Kpler data — evidence of recovery — even as the tanker-turnaround claims pushed prices the other way. The result is a wide, nervous $80–$100 band that lurches on each new report.

The tails are fat. Goldman Sachs has warned Brent could reach $120 if disruptions persist through both Hormuz and the Red Sea; one analysis pegs a three-month disruption at a $114 monthly average. The infrastructure is adjusting to a long war: ADNOC has bought five VLCCs for about $590 million to expand its fleet as tanker supply tightens, and Saudi Arabia says it has assembled a 14-country coalition to secure sea routes. Middle-distillate markets are separately tight — Russia extended its diesel export ban to September 1 after Ukrainian drone strikes on its refineries, and the gasoil crack is near record highs above $70.

Strip it down: a chokepoint two-thirds closed, a market that cannot see clearly through it, and a war that reprices every barrel each time a drone or a tanker makes the wire. That is the machine now setting the price of a gallon in America.

Here is a number that should mean something to anyone who keeps declaring the theater dead. On Thursday, *Spider-Man: Brand New Day* opened in India to roughly ₹72.44 crore gross — about ₹60.60 crore net — the biggest opening day for a Hollywood film in Indian history. It beat *Avengers: Endgame*, the seven-year benchmark. It beat *Pathaan*. It nearly doubled *No Way Home*, its own predecessor. Across 17,250 shows, occupancy ran at 72.3 percent — a full house, morning to midnight.

Destin Daniel Cretton directs; Tom Holland and Zendaya headline the 38th film in the Marvel Cinematic Universe and the fourth solo Spider-Man. It is projected to clear ₹225 crore over an extended weekend, and trade watchers are openly speculating it could become the first non-Indian film to cross ₹500 crore in the country — territory only *Avatar: The Way of Water* has approached.

The detail I keep turning over is the one comparison the trade press buried. On BookMyShow, day-one bookings for *Brand New Day* outran Christopher Nolan's *The Odyssey* — the film Holland *also* stars in, the one shot on 70mm with analogue projection and pitched as an event for people who take cinema seriously. The web-slinger beat the epic. The same actor, two theories of what a movie is for, and the crowd chose the one with the joke and the mask.

I don't read that as a defeat for ambition. I read it as a fact about what a full room wants right now, in a summer where every other headline is a strait or a border or a drone. Sixty thousand people crossed a fence into Ceuta this week; sixty thousand more, give or take, packed Indian multiplexes to watch a kid from Queens have a bad day and win anyway. Both are crowds voting with their bodies. One vote is desperate. The other is a small, deliberate act of choosing delight in a bad month — which is not nothing, and which the numbers say the world is very much still willing to pay for.

I want to be fair to the thing before I take it apart. Hamas said a word on Thursday it had never said. It communicated, through mediators, that it would decommission its weapons. After nine months of deadlock, a group that had never once agreed to a mechanism for disarmament agreed to one. That is a real event. Cynicism that cannot admit it is just a different flavor of laziness.

Now: everything else.

An agreement is a set of obligations that bind you against your will — that is the entire point of one. You sign because signing costs you the freedom to change your mind. By that standard, what was announced Thursday is not an agreement. It is an agreement to produce an agreement, with the binding parts deferred and the dates left blank on purpose.

Count the escape hatches. No timeline for disarmament — officials said so directly. The police hand over weapons in two weeks, but the *heavy* weapons and the tunnels come "later," in a window an official floated as 200 to 350 days. You do not proceed from phase to phase until a verification committee blesses the last one. Israeli withdrawal is pegged to progress on a timetable "still to be finalized." And the detailed conditions — the actual contract — will be "published within two weeks."

Every one of those clauses is a door that opens outward. A 350-day process is not a plan; it is eleven months of opportunities to declare the other side in breach. A verification committee is not a safeguard; it is a veto with a nicer name. "To be finalized" is where deals go to quietly not happen. We have watched this exact structure before — the phased, conditions-based, verification-gated framework — and we know its failure mode, because it has a name in the diplomatic trade. The name is "implementation." It is where the announcement goes after the cameras leave.

I am not saying it will fail. I am saying that anyone calling it a success today is grading the press release, not the peace. The Israeli source who called the withdrawal a "trick," the former army chief who warned a deal that leaves Hamas standing is a "complete failure," the Islamic Jihad statement calling the whole thing "inaccurate" — these are not spoilsports. They are the people who will have to actually execute this, telling you the ink is not dry because there is no ink.

Here is my test, and I will hold this paper to it. In two weeks, the conditions are supposed to be published. If they arrive with real dates, real enforcement, and a real cost for walking away, I will write the column admitting I was too sour. If they arrive as another framework to be refined — another round of "delicate negotiations" producing an "understanding" — then Thursday was theater, and the applause was for a curtain, not a play.

Declare peace all you like. Just don't ask me to celebrate a treaty that forgot to include the date it takes effect.

The AI press this month wrote its usual liturgy about the top of the stack: Anthropic's Claude Opus 5, OpenAI's flagship Sol, the benchmark scores, the frontier. And it mostly missed the actual news, which happened at the bottom.

On July 30, OpenAI cut the price of GPT-5.6 Luna — its cheapest tier — by 80 percent, to twenty cents per million input tokens and $1.20 per million output. It cut the mid-tier, Terra, by 20 percent, to $2 and $12. The flagship's price didn't move. By one accounting, Luna now undercuts Anthropic's cheapest published model by roughly five times on input. This is not a feature release. It is a price war, and price wars are how technologies stop being events and start being infrastructure.

Here is why the floor matters more than the ceiling. The flagship model is a status object — you reach for it on the hard, high-stakes, one-off problem, and you notice the bill. The cheap model is what you build *into* things: the thing that runs a million times a day inside a workflow nobody thinks about, cleaning data, triaging tickets, tagging documents, doing the boring middle of a business at a cost that finally rounds to zero. OpenAI said as much — Luna makes "high-volume work economical at much greater scale." That sentence is doing more to reshape the economy than any benchmark on Sol.

And notice how the two companies now converge on the unglamorous parts. Both sell a "Fast mode" at 2.5x speed for 2x price under the same name. Both price region-pinned inference at a 1.1x premium — OpenAI's data-residency uplift, Anthropic's US-only rate, arriving at the same number. When rivals' rate cards start rhyming down to the data-residency surcharge, you are no longer watching a technology race. You are watching a commodity market form, complete with tiers, discounts, and the beginnings of a utility's pricing psychology.

There is a tell buried in the accounting, too, and it is the most interesting thing anyone published this week. OpenAI says the model helped make *itself* cheaper: Sol, running inside Codex, rewrote the company's production GPU kernels and redesigned its own speculative-decoding draft model across hundreds of experiments, cutting serving costs by around 20 percent and lifting token efficiency more than 15. The frontier model's most economically consequential act was not answering a hard question. It was lowering the price of the model underneath it.

That is the loop worth watching. Not "which model is smartest," a question that will be re-answered every eight weeks and matter less each time. The question is how cheap the *floor* gets, and how fast, because the floor is what the world is quietly being rebuilt on top of. Twenty cents a million tokens is not a headline. It is a foundation. Cover the foundation.