I cancelled my Claude Code sub. I give up.

The frustration is valid but this is a dependency problem, not a tool problem. If one provider's pricing change breaks your entire workflow, you built a vendor lock-in not a system. Provider-agnostic constraint files survive any subscription change.

Isn't this fair? I mean, for the people using Anthropic's products this is ideal. If you are using different tools out of their ecosystem, pay the bill at API level.

$300 on super @grok now?? I'm working to integrate it into Trifecta right now! Oh, I also got ACP registry to start working along with Devin, Gemini, and Hermes!

I fired Claude after refusing to work on a React project because it's too dangerous. There was a lot leading up to that final straw.

You CAN run claude on anything you want, but you have to pay for it through the API. It has always been clear that the subscription is heavily subsidized to promote their apps. I don't understand why people still get outraged at this. It's like asking them to give their product for Free at this point.

A full crash out is completely justified. They literally rugpulled the entire developer community overnight 😤

Fucking wild what OpenAI just did. Game changer.

I’m also about to give up very soon. 99%

Where is TLDR ? 45 min video is too long to know the reason 😄

Hearing this for the nth time now

valid crash out, fuck anthropic

So, they are supposed to subsidize your 3rd party endeavor ?

@DanielleFong I kept my Claude Code sub because someone needs to stick around to give them bad training data.

Dude has been a bitch ranting for days that Anthropic is not subsidizing his little project anymore. Throwing a tantrum on Twitter and asking other morons to follow him cause he’s offended.

I'm afraid the economy of it just prevents them to subsidise all usage types like they do for CC Anthropic is a for-profit company, not a some-day-(never)-profit company like openAI Love your content tho

We had two sites cancel their Claude Code subs. One had ~80 seats another ~100. Today. And mind you, they haven't been hit with this, they just read about it. Wait until they actually make the change. The funny thing is that neither of them are doing anything "agentic". They just want a friendly front end (which is what our app does) to put the AI and their data together. If they don't change their policy to make a distinction between "mass bots" and custom Claude Code processes, OpenAI and Copilot are going to eat their lunch. If they think companies are going to mass adopt Cowork or Claude Desktop they are delusional.

I feel this. Half the job now is figuring out which subscription is going to betray the agents mid-run.

But Mythos found one curl bug that the devs weren’t worried about and said they’d just get to it in next scheduled release in 2 months. How ever are you going to manage without a relabelled opus when that comes out?

Good video as always. One important nuance: being on Enterprise + token-based billing doesn’t automatically unlock the broader ecosystem. Even if your company pays for tokens, using tools like @opencode still requires separate enablement beyond the @claudeai Enterprise plan.

This is even worse than Claude Code. 😂 How is that even possible? Pay 300 bucks to build stuff with our terrible model no one wants to use. What a bargain!

believe me. i am a huge openclaw guy and my heart broke when they relegated it to API billing but still …

Urgh I hate that I feel like I’m making excuses for CC. I’m still using both max subs but CC is just like a bad employee that I can’t let go.

codex has been cooking lately and claude is just pissing users off lmao

Jeez.. what a crybaby, I have my fair share of Claude criticism but for the past 2 days my feed is filled with your desperate attempts to move users from Claude to your shitty alternative. Isn't this your 50th time cancelling already.

Developers unironically saying “Can’t the users just use the CLI/call the API?” used to be the punchline of a joke about why Product Managers/Owners are necessary. It wasn’t supposed to actually play out exactly like the joke at this scale

so wait, if i use claude -p command in my own local agentic workflow that's now a separate usage?

Qwen code is the best atm and you can just run multiple models on openrouter for the same price. Important thing is to always adapt to your situation rather than complain about a tool

the one good thing that come out of this is I've moved from claude code to t3 + chatgiptty and it's like I've seen the light 🙏 those terminal code harness days can GTFO

I canceled it yesterday. Purchased it yesterday. Canceled within 45 minutes of using it for code. I felt like I was at a slots machine. $20. 90% used in less than an hour.

Removing 'claude -p' as an escape hatch for their terrible harness is just pure tyranny They know their harness sucks so their only "solution" is to force people into it, lmao

I think I found a decent workaround for this. Basically keeps the `claude -p` feel, but runs through an already-open Claude Code session instead. Wrote it up here:

our team is holding on with 5 max accounts in parallel but the speed/cost calculus stopped making sense for solo devs this week. codex is no longer the meme option

I'm using opencode with my minimax plan on my tasks orchestrator tool

To use your restaurant analogy (28:20): what you did was open another store next to them and tell people with a buffet ticket that they can bring Anthropic’s food to your place, and you don’t charge a dime. Now do you know why no buffet allows people to take food outside?

Are you gonna sub to Grok SuperHeavy now?