Anthropic can't stop winning
Monday, Nov 2 · 17 min · about 6 min to read
Theo-brain watched all 17 min
Anthropic's $200/month Claude plan is now called Ultra, at the same price, and verified accounts no longer have the 5-hour usage limit. The catch: apps using Anthropic's agent SDK, like Theo's T3 Code, get a smaller 5-hour allowance than Claude Code itself. Theo got informal permission to ship a fix so T3 Code users get the same usage as in Claude Code.
Briefing is enoughThe notes give you what you need, so skip the video unless you want to hear Theo's full reasoning.
Tap any time to jump the video there.
Key takeaways
Ultra is the old $200/month Max 20x plan at the same price. The real change is that it drops the 5-hour limit, a cap on how much you can use in any 5-hour window. That only applies to accounts Anthropic has verified.
more money. It's the exact same plan as the 20X was before, that is the $200 a month plan, but they made some subtle changes to it that are really worth noting. In particular, they removed the 5-hour limit. This is just a great
Oh, for verified ultra accounts.
The 5-hour limit exists because of computing capacity, not cost. Usage peaks on weekday mornings, Pacific time. Anthropic needs spare GPUs then to sell to big companies, so the limit stops one user from burning a big share of their week in one window.
5-hour limits were introduced as a way to prevent a subscription user from taking more than a certain percentage of their weekly sub in any one of these work windows. If you broke up this chart
Apps that use your Claude subscription must either run the Claude Code command-line tool directly or use Anthropic's agent SDK (a toolkit for building agents). The SDK route cuts your 5-hour allowance by 50%. One user found T3 Code uses about 1.8x more of the 5-hour window per token than the CLI.
in your apps, you either have to have your app just call the Claude Code CLI directly with like the terminal visible, or you have to use the agent SDK, which means you're going to be hit with this 50% decrease in the 5-hour limits.
like T3 Code. This user pointed out that T3 Code uses around 1.8 times more of the 5-hour window per token than the interactive CLI does. Even {dash} P does
While filming, Theo got informal permission from Anthropic to merge a change to T3 Code. Once that release ships, T3 Code users should get the same 5-hour and weekly usage as in Claude Code directly.
see this 2 minutes ago. Yeah, I got semi-solid permission to merge this which should, hypothetically, fingers crossed, get around the 5-hour limit changes that affect T3 code so that you will get the same usage for the 5-hour
For most people, the $200 plan is more than enough. Theo says that with his heavy automated workloads turned off, he struggled to use even half of one Ultra account.
realistic dev work. Hell, even some heavier stuff would be totally fine in Ultra. When I turned off all my hill climbs, I struggled to even do half an account with Ultra right now, if I'm being real. Max is also quite generous
Skills in this one.
Routing Claude usage through a CLI proxy
- What it is
- Theo sends all his AI usage through a CLI proxy, a middle layer between his apps and his Claude accounts. It already sidesteps the agent SDK penalty and handles juggling multiple accounts.
- Why Theo cares
- He never hit the T3 Code limit problem because the proxy handled it. He says he has used this setup for at least four months with no issues. He also warns that he can't be held responsible if it gets an account banned.
- How to try it
- Find the setup Theo shares in T3 Code's FAQ and have your coding agent read and explain it.
- Weigh the account-ban risk he flags before using it on a business-critical account.
- If you run several accounts, don't manage them inside T3 Code directly, as he advises.
said, I didn't notice these problems because I've been routing through CLI proxy anyways, which already applied things like this to prevent this issue. So, again, if you wanted to have multiple accounts or manage all of that, I don't recommend doing it in T3 code
Advice
To see whether your account is verified, go to Settings, then Usage. If there is no current session bar, you're verified. Verifying your phone number at checkout does not count.
you need to do. Verifying your phone number at checkout is a separate step and does not remove this limit. To check, go to settings usage. If you don't see a current session bar, your account is verified. Yeah, not great. It
Hot takes
The $200 subscriptions make the AI labs little or no money. They exist to get developers hooked on huge usage so their employers end up paying full API prices.
His reasoning: He says you can get as much as $12,000 a month of usage for $200. People who get used to that expect the same at work, where companies pay per use through the API.
it's also worth noting that these $200 subs don't make the labs much, if any, money. The point's to get us addicted to using 12 grand of tokens a month so that when we go to our employer, we're expecting to spend a similar amount there and they'll have to pay those API
Removing the 5-hour limit only from the $200 plan is backwards. That limit hurts the cheaper plans much more.
because remember, they're only removing this from the $200 plan, even though those 5-hour limits tend to hurt a lot more on the cheaper plans. Ah, there's a
Anthropic is unusually vague about what verification even means. Theo dislikes this but thinks they can't explain it.
His reasoning: Labs, including Chinese ones, buy subscriptions and burn through them to collect responses for distillation, which means training their own models on Claude's output. No 5-hour limit makes that easier, so the checks have to stay opaque.
removed. Yeah, I don't love this. I specifically don't love the lack of transparency here. They have been truly like legendarily vague about what verification even is. They say they
and then use that for distillation. The removal of the 5-hour limit does make it slightly easier to do that, which is why they want to enforce things more strictly this way, but it's also just
How Theo thought about this
To explain why the limits exist, Theo checked T3 Code's own usage data. He found a 2x gap between the busiest and quietest hours of a normal weekday. That made the case that the limits are about scarce computing power at peak times, not about cost.
There is a 2x gap in how much prompting is being done from the lowest lows and the highest highs just in normal variance of a given weekday. I bring
When users reported that limits burned faster in T3 Code, he didn't take the report at face value. He ran his own numbers to confirm it, then worked with Anthropic directly before choosing to ship a fix.
reporting it to us in T3 Code. I did my own calculations to confirm this, and it is indeed true. I've been working with Anthropic on this because I was hoping it would get fixed. I patiently waited,
Sign in to mark what you've watched and keep notes.
Have a question about this one? Ask about this video →
Theo Daily is unofficial and not affiliated with Theo or t3.gg. Written by Claude Opus 5.5 from each video's transcript.
