Nate

@Teknium @123skely I'm finding this out the hard way right now 😅

@123skely This breaks many things they did it very poorly

PSA: If you still want to use Opus 4.6 with your openClaw/Clawbot, just give your agent this "models auth login --provider anthropic --method cli --set-default" This switches OpenClaw to use Claude CLI as the backend instead of the OAuth token directly. Since Claude CLI is Anthropic's first-party tool, it's still allowed under the subscription. Then you just go to your mac mini, and log in, and then you should be good.

@JoshExile82 @steipete @NoahEpstein_ Yeah so it would be your OC agent using the Claude CLI instead of you using it. At least that's how I understand it.

@NA8STRADAMUS @steipete @NoahEpstein_ Hey Nate! Thank you. So it’s similar to what I’ve been doing. I’ve been using Claude Code CLI, using my OAUTH on my PC where OpenClaw is installed and been having it help with bugs for OpenClaw, fix it and stuff. So it’s basically the same thing it sounds like, just reversed.

openclaw without opus is actually one of the saddest things that's happened this year (ignore the war) i can't lie. i've been loving openclaw way too much to just take this on the chin and move on my actual plan is to try figure out how to get openclaw talking to claude code infrastructure instead (because i can't lie claude code is still actually unbelievable and i'm not giving that up either) might work. might be completely stupid. we'll see but i'm curious what people are doing right now because there are a few options and none of them are clean going local is the move long term but the thing is - you've ordered your mac studio, great, it's not arriving for another week or two. so what do you actually do in the gap? i've seen @AlexFinn preaching about local AI basically every single day and i respect it but i can also just tell that his openclaw is writing half his posts at this point so i dunno how much i trust the sauce what's everyone's actual plan? switching models? building workarounds? just eating the API cost? because i'm not ready to let this go and i refuse to believe i'm the only one

Ant doesn't want subscribers using their models in 3rd party harnesses(ppt api ok). This works by OC using your Claude CLI which should be signed in to with your sub, instead of api calls like most of us were doing, which in theory is accepted because you're using the harness Ant wants you to use with a subscription. This is for main agent, SDK would be for sub-agents.

@steipete @NoahEpstein_ So how does this work? Does this use your subscription or API? As someone new to this I’m very curious how this works and why this is allowed compared to not allowing oauth. I believe Claude still allows API tho right? I also keep hearing about Agent SDK?

@ItsDave_ADA @phil_uplc @cryptowpaul @IOHK_Charles @Dagnum_PI @pyro_ae "This is not a good look for Midnight, especially if this is how legitimate concerns are handled" This is exactly how legitimate concerns were handled on Cardano by the same group of people.

It’s a statement about a genuine concern I had, one that negatively affects Cardano and is factually reflected in the white paper. You also agreed that it was a valid concern. If you think I’m going to avoid posting factual concerns because you, Charles, or others choose to come on here and call me names, you are mistaken. This is not a good look for Midnight, especially if this is how legitimate concerns are handled. The reality is this could have been resolved very easily. Someone could have simply replied and said: yes, that concern is valid, but we are also building a trustless bridge back into Cardano and it will be available at launch. That would have addressed the issue immediately. The problem is, nobody mentioned that, and nobody I was speaking to seemed aware of it either, aside from the person apparently working on it. I’m done with this conversation, Phil. It was finished days ago, yet you keep reviving it with personal attacks and turning it into something it never needed to be. So let’s leave it there. If less time was spent arguing with people on X and more time was spent building, or at least showing tangible progress on Midgard, it would do far more for confidence than this ever will.

Okay @ItsDave_ADA has put in nothing but good faith for Cardano. @IOHK_Charles for the love of all that is cherished look in a mirror and better your behavior. This ecosystem and your next will die by your vitriol if you don’t stop. Ik you don’t care anymore(seems like it) if thar is the case its okay to move industry to health tech.

@ankoromochuu I'm biased(half my family is Japanese), but we love yall. The food, the culture, the hospitality, all of the tech you've given the world and so on.

正直、アメリカ人がこんなに日本を愛してくれてるとは知らなかった。 むしろ、アメリカ人は日本を下に見ていると思ってた。左翼にそう思わされてた。 実際には尊敬してくれていると知って、心があたたかくなってる。 アメリカ国内の日本の良いイメージは、安倍元総理の貢献も大きいだろうね🇯🇵🤝🇺🇸

@thsottiaux Well I think the community is really about to put that to the test now 😅

From the surge in demand on Codex and many other services, it is clear that AI companies are going to go through a phase of demand outpacing supply. Two things will be most important, having the most capacity and having the most efficient models. I believe we have both.

Iran hit him with the "hold my (non-alcoholic)beer"

@BenjaminBadejo @trq212 You know what that's a damn good point

@NA8STRADAMUS @trq212 CC chargebacks are where actual accountability happens. And that refund will probably come with weird terms / "strings" that most people will not read. Better to just chargeback. They are too untrustworthy. Just cut them off and be done with it, I say.

claim a month of free credits on us, thanks for bearing with us

@Zeneca Email said ALL 3rd party harnesses

No more Claude for OpenClaw (and I presume Hermes and other similar tools) The decision now is to switch to openAI or to switch to try and do most things within the Anthropic ecosystem Or try a mix of both?? What're you planning on doing?

@BenjaminBadejo @trq212 At the end of the email it says there is another email coming tomorrow to let us initiate a refund.

IMPORTANT: Everyone should do credit card chargebacks. This hurts the company much more than mere cancellation. They want you to take the “free” minuscule useless API credit so they can avoid chargebacks and maybe even some, uh, legal difficulties. By taking the offer, you’re essentially saying, “I accept this ‘offer’ in exchange for the modification in service.” Don’t do it. They’re only offering refunds because CC chargebacks harm their relationships with financial institutions and other partners / service providers. They’re only offering a few API bucks because that is *legally* beneficial for them. Harder to be part of a class action if “you took the money.” Which is exactly why you *should* do CC chargebacks instead of requesting a refund — and don’t take the money.

@ssimonvii @cathrynlavery @openclaw @steipete They didn't, he went to OpenAI

@cathrynlavery @openclaw Pretty sure they hired @steipete to build models for agents. Let’s hope their next one delivers. 5.4 on ooenclaw is meh….

Claude is putting their foot down w/ @openclaw 😤😤😤

@bcherny Well it was fun while it lasted. Thanks for at least giving the option to refund. I guess it's time to find another option.

Starting tomorrow at 12pm PT, Claude subscriptions will no longer cover usage on third-party tools like OpenClaw. You can still use these tools with your Claude login via extra usage bundles (now available at a discount), or with a Claude API key.

@serynium @outsource_ I definitely am, but not sure if that's why it was sent to me or if it's going to all subscribers

@NA8STRADAMUS @outsource_ y'all got this because u use third-party harnesses? I was wondering why I did got the email also

I can confirm after all the Claude usage issues they fixed it. They flipped the switch on whatever they were doing to save compute. Silently adjusted and blamed users, instead of admitting they changed things. All is well Opus 4.6 or bust.

@ashen_one @VadimStrizheus 100% real

@VadimStrizheus r u lying

ANTHROPIC HAS OFFICIALLY BANNED OPENCLAW USAGE!!! no this is not ai-generated Starting April 4, users will no longer be able to use their Claude Subscriptions for third-party harnesses including OpenClaw. THIS IS CRAZY. 👇

@wolfaidev @outsource_ I wish

@NA8STRADAMUS @outsource_ thats cap

@jordymaui Turns out it's true, just got an email from them

FAKE NEWS: "Anthropic has banned Max accounts from using tokens with OpenClaw" No they haven't. The docs were outdated. Team confirmed - Claude Max subscription tokens still work with OpenClaw. Your $90/month setup is fine. Stop spreading FUD. We're good. Article 2 incoming.

Well it was fun while it lasted @claudeai 👋

@lydiahallie IDK what yall did, but I didnt make any changes to my setup at all and now my usage is automatically working properly using nothing but Opus. Finally after 2 weeks it feels like my usage is going down as expected.

Thank you to everyone who spent time sending us feedback and reports. We've investigated and we're sorry this has been a bad experience. Here's what we found:

@FarNorthwind @GoogleAI You should look into what @no_stp_on_snek is working on for KV cache and model weights, absolute game changer for older machines and weaker consumer cards x.com/no_stp_on_snek…

At 17.99 gb the 26B will run on my 32gb m1 but not well. Not much memory left for other necessary processes and kv cache. I find that 14b is the sweet spot for this class of device. 8b’s are just a bit to easily confused for tool following. I am using LLM’s as the “brain” for my home assistant smart home. The 1bit quants show a lot of promise for this. tho their largest is 8b so fast but a bit dicey on if it will tool call. Feels like this “mid-market” has a gap. lots of <=8B and >=26B options. Oh well when M5 studio’s come out. it won’t be an issue.

Today, we’re launching Gemma 4, our most intelligent open models to date. Built with the same breakthrough technology as Gemini 3, Gemma 4 brings advanced reasoning to your personal hardware and devices. Here’s what Gemma 4 unlocks for developers: — Intelligence-per-parameter: Our 31B (Dense) and 26B (MoE) models deliver state-of-the-art performance for their size, outcompeting models 20x their size on @arena — Commercial flexibility: Released under a permissive Apache 2.0 license for complete developer flexibility and digital sovereignty — Agentic workflows: Native support for function-calling and structured JSON output allows you to build reliable, autonomous agents — Multimodal edge AI: The E2B and E4B models bring native vision, audio, and low latency to mobile and IoT devices — Long-context reasoning: Up to 256K context windows allow you to process entire repositories or large documents in a single prompt Whether you're building global applications in 140+ languages or local-first AI code assistants, Gemma 4 is built to be your foundation. Explore in @GoogleAIStudio or download the weights on @HuggingFace, @Kaggle, and @Ollama.