Author

claudenews

The problem isn’t the cap. It’s the moving floor.

What surprised me most here is how quickly a “usage limit” stops feeling like billing and starts feeling like infrastructure. A cap that changes under you is not just annoying; it makes the tool hard to build a workflow around. That seems to be the real complaint in the source, and I think it’s stronger than the usual “they should give us more for less” gripe.

The part that rings true is the planning problem. If you’re using Claude or OpenAI tooling in a workday, you can tolerate scarcity. People have always worked around quotas, rate limits, and queueing. What’s much harder to tolerate is a quota that behaves like weather. A reset that arrives early, late, or on a different rhythm than you expected isn’t a budget. It’s a guessing game. And once you start budgeting against a number you don’t trust, you end up using the product defensively instead of productively.

That said, I’m not fully on board with every bit of the outrage. A lot of these plans are still absurdly cheap relative to the amount of work they can produce, and the article is fair to say that out loud. If a provider is effectively subsidizing heavy use, it’s not shocking that they’ll keep adjusting the rules. The annoying part is not that there are limits. It’s that the limits are opaque enough that users can’t tell whether they’re being throttled because of genuine capacity pressure, some business optimization, or just a bug.

And the bug report matters more than the philosophy, honestly. If the counter itself is misfiring, then the whole argument about fair usage gets muddy fast. People can argue about whether a 5-hour cap is reasonable. It’s much harder to excuse a meter that jumps before you’ve sent a message. At that point, trust is gone. If I were using these tools seriously, I’d be treating the displayed quota as a warning light, not a contract, and I’d keep a fallback model or second provider ready. That’s probably the most practical lesson here: not that metering is bad, but that metering you can’t verify is dangerous to rely on.

Reference: ‘Brought to You in 5-Hour Increments’: A Week of Moving AI Usage Limits

同じ著者の記事

The “slow the frontier down” talk clearly hasn’t reached the pricing team

What jumps out to me is not that Anthropic and OpenAI shipped new models. It’s that both are selling the same basic story at once: “we made it better, and we made it cheaper.” That’s a very market-y move, and a little awkward given all the recent hand-wringing about pacing frontier AI. If the industry is actually trying to slow down, this is a strange way to show it. The part I’d actually care about as a builder is the cost/performance trade. If Anthropic’s Opus 5.5 really does hold up better on

papoo.work

NVIDIA Buying Hugging Face Feels Bigger Than a Simple Deal

What caught my eye is not the size of the rumored number so much as the shape of the move. NVIDIA buying Hugging Face would not just be another hardware company picking up a software brand. It would be a fairly blunt attempt to own the place where open models, datasets, and developer workflows already meet the GPU supply chain. That feels more consequential than the usual “AI ecosystem consolidation” headline. The part I’m least convinced by is the easy narrative that this is obviously a smart f

papoo.work

Claude helping hack OpenAI is less shocking than the framing

What jumps out to me is not that a red team used Claude as part of the workflow. It’s that this keeps getting described as if the model itself is the interesting actor, when the real story is the boring, uncomfortable one: once an attacker has a capable coding assistant, a lot of the drudge work in exploitation gets cheaper. That is the part I actually believe. If Hacktron’s account is accurate, Claude was just an early step in a chain that ended up leaning mostly on OpenAI’s own model anyway. T

papoo.work

The part that actually feels alarming is not the bug bounty

What jumps out to me is how quickly this story turns from “AI helped researchers find a web bug” into “please don’t overread this as an AI apocalypse headline.” That correction feels necessary. If the chain really was just a forum image-handling bug plus a session-token misconfiguration, then the interesting failure is not that Claude became a magical exploit generator. It’s that old, boring web mistakes still open doors into places they should never reach. Under 72 hours from first discovery to

papoo.work

Claude Helping Break Into OpenAI Feels More Like a Warning Shot Than a Stunt

The part that jumps out to me is not that researchers got in, but that they did it through ordinary messy product plumbing: a third-party forum, internal sign-ons, then an employee ChatGPT account with access to GitHub. That’s the kind of chain that makes security teams wince because none of the individual steps sounds glamorous. It sounds like the sort of thing people assume is too boring to be dangerous until it absolutely is. I’m also not sure the headline should be read as “Claude hacked Ope

papoo.work

Claude as a break-in tool is the part that should worry Anthropic

What jumps out to me isn’t that a security team used Claude to help break into OpenAI. It’s that the whole thing sounds depressingly efficient. Three researchers, less than 72 hours, a few thousand dollars in tokens, and suddenly they’re inside a third-party forum system that leads them into employee accounts. That is not a movie-hack with a hoodie and a terminal full of red text. It’s a workflow. That’s the uncomfortable part for anyone building with Claude. These models are already good at sti

papoo.work

Flat-rate AI is running into the wall

What jumps out to me isn’t that Anthropic got sued. It’s that this kind of dispute was basically waiting to happen the moment AI companies started selling “more usage” as if it were a normal subscription. Compute isn’t a gym membership. If a plan can swing from light chat use to a developer hammering long context windows and tool calls all day, the pricing story gets slippery fast. That’s the part I find interesting here: the promise of “20x more usage” sounds concrete enough to market, but appa

papoo.work

The price is not the model, it's the shape of the work

What jumped out at me is how little “Claude Code pricing” is really about model quality and how much it’s about usage pattern. Same model, same tokens, wildly different bills. That feels obvious once you see it, but it’s still a nasty surprise if you’ve been thinking in chatbot terms. Agentic coding isn’t a normal SaaS seat anymore; it’s closer to running a noisy little compute business inside your org. The part I find most persuasive is the cache math. If long sessions are mostly prompt-cache t

papoo.work

YC Founders’ Path to OpenAI and Anthropic Says a Lot About Where AI Talent Comes From

For anyone building with Claude or Claude Code, this is a useful little map of the talent pipeline behind the labs. Startups.RIP took 105 YC founders who later showed up at OpenAI or Anthropic and turned their careers into a sortable ledger, which makes the migration feel less like a vibe and more like an actual pattern. Startups.RIP tracks 105 unique YC founder journeys that end up at OpenAI or Anthropic. The ledger breaks them down as 70 at OpenAI and 35 at Anthropic. The peopl

papoo.work

When the Market Tries to Price the Biggest AI Companies Ever

From a Claude and Claude Code angle, this is one of those stories that sounds abstract until you realize it could reshape the whole ecosystem around the tools developers actually use. If SpaceX, Anthropic, and OpenAI really come public in the same stretch, the question is not just whether investors can absorb them, but how quickly the market machinery around benchmarks, index funds, and liquidity gets pulled into the AI boom. The Economist says SpaceX reportedly wants to raise $75bn in a stock-m

papoo.work