Saw this coming eventually. $20/month for autonomous agents running 24/7 was clearly not sustainable at API pricing. The part that's surprising is there's still no official announcement - just a quiet page edit.
Choosing to do it quietly instead of letting everyone know is actually not that surprising.
And yet they're very aware that Hacker News, etc exists and so the awareness and backlash would be instant. It's like they want to get a lower rating from the community. Maybe that's their solution for the resource issue: make enough people mad so they abandon their subscriptions.
If true, very strange change when Codex (at both 20 & 100) is a much, much better deal for a model much better at most coding tasks, with way more usage even with the /fast mode enabled. Is losing most non-enterprise customers the right move for them?
Equally, will offering a presumably unprofitably large quota of Codex tokens at $20 to retain non-enterprise customers turn out to be the right move for OpenAI? Would not be surprised to see OpenAI follow suit. Or perhaps OpenAI's LLMs are just so more compute efficient that they can actually offer that sustainably...
Feels to me it's a battle between who has the most compute. OpenAI does not seem to be struggling with their x2 usage on the new 100 Plan, which is very close to unlimited usage with the best performing model on the highest reasoning setting. Not mentioning the resets every 1 million customers, or the other generous usage multipliers last months. Meanwhile Anthropic seems to be desperately trying to cut down on inference with their changes to reasoning effort and more lately, so they might be focusing on what they consider to be more valuable customers for their long-term strategy. The 20 plan with Opus had gotten so bad on CC they might've just pulled the plug to stop people from complaining about usage limits. If OpenAI can burn money longer and capture the market from the bottom, I think they'd win in the long run.
Makes sense. It is over for the little guy - home enthusiasts and vibe coders. Too many of them saturating resources for Max users. IF you cannot afford few hundred dollars subscription go out and breathe fresh air. But if you can, watch where the ball is rolling - few thousand dollars subscriptions and even less programmers.
Hear HN tell of it, Claude pays for itself 3× over. Something tells me congitively it's making us misjudge how productive it's making us. It's clearly massively increasing output, but did the market already soak up all that productivity and now it's not compensated? If your salary is 50k. And Claude makes you 2x as productive, why aren't you earning 100k? Why is it anyone can't afford $200/mo if it's truely increasin…
Because most people work for someone else and don't decide their own salaries. It's not doubling productivity, but even a 10-20% boost to productivity for a team of engineers means that, as a business, even $1k per month per seat is perfectly acceptable. For consumers and hobbyists that basically kills access.
I had a similar ride, but disagree with your conclusion. Opus 4.7 is so incredibly powerful from my experience, that nothing else really matters and I think at Anthropic they know it. People will pay a lot for access to this model.
I've had completely the opposite experience. I've asked for it to research things and it's just told me to "paste xyz into google". Just now I revisited a chat that's 5 days old and asked it to check again (because what I was looking for might have changed), and it said "no".
It's funny how experiences can be so different. I wonder if this comes down to context. My interactions so far were fairly high-level and in some cases it having a strong opinion was actually super beneficial to the outcome. To me it seemed opinionated, but in a very good way. I can see how this could backfire though and have heard similar reports.
I think that Anthropic has capacity problems. They went all in on acquiring new customers but now they don't have enough capacity to both serve users and train new models, so they are trying to limit user usages. It is pure speculation of course, but I don't have any other explanations on the stuff they are pulling in the last 2 months.
Yeah, this is pretty clearly what's going on, but I wish they'd be more transparent about it. Funneling compute to Mythos and Design, while auto-setting effort levels lower and removing user control of extended thinking. I don't think the need to shuffle compute around is unique to Anthropic, though. I suspect it's part of why Sora got killed. And everyone's having uptime issues. Are we reaching the limits of the available compute?
It was obvious, which is why their earlier decision to attempt to restrict 3p access was hugely unpopular. They're now trying to walk that back but on the condition that those users need to be on the higher plan.
Now they're hurting their popularity with those who actually don't mind using Claude Code. I've been quietly swallowing a lot of performance degradation over the past few weeks as I get the resource crunch, but I'm definitely not going back to copy-pasting between browser and editor. And I have no intention of upgrading to Max or doing per token usage.
It seems they were losing money on pro subscriptions with code so if people stop using them it's actually ok for them.
They have now moved to be enterprise providers and don't need the cheap pro users as loss leaders anymore.
Everyone that is upset about this should take note: you are not a (coding) customer at $20/mo. Their coding customers spend thousands per month (week!) on claude and it's growing faster than they can keep up with (source: I'm one of them, and I know many other like me. We're budgeting 10-20% of engineering salary spend on tokens). It sucks to no longer be able to code on the cheap anymore, but don't fool yourself into thinking you have any leverage here.
Anthropic clearly doesn't understand that customers see their brand as "Claude", Google's brand as "Gemini" and OpenAI's brand as "ChatGPT." They have so many plans and exclusions that they risk customer confusion. I was surprised when I was pay $200/month for Claude Code, finding it super helpful, and then I had to pay separately to get API access for an experiment. Why are so many parts of "Claude" separate from ea…
That also was really opaque to me RE: API access. I initially thought at $200/month I could get whatever I needed. I eventually set up a OpenAI API with a few bucks to try what I wanted to.