Andrew Sperazza 🇺🇸

@bridgemindai @predict_addict Unfortunately same view here.

Today is the LAST DAY Fable 5 is on Claude subscriptions. So let me be direct with Anthropic: GPT 5.6 is way better than Opus 4.8. Without Fable 5, there is nothing on your subscription worth paying for. I have 4 Claude Max subscriptions. $800 a month. If Fable 5 is not brought back, I cancel all of them immediately. Your move.

@trq212 You can drop opus just go with fable. Capacity solved

I've heard a lot of questions about Fable's availability on subscription plans. While it will come off subscriptions after July 7th, we aim to restore Fable as a standard part of our subscriptions as soon as capacity allows, as we mentioned in our original blog post.

@morganlinton that is good advice, didn't even seriously consider it , but will try.

I'm on the $100/mo Claude Max plan and I have been running Fable 5, pretty much non-stop since it was released, and can't seem to hit any usage limits. About 95% of what I do from a coding perspective, only requires Fable 5 Low and Medium. (and note, low and medium are out-performing Opus 4.8 and GPT 5.5 High and xHigh in a lot of benchmarks) And I have been blown away with how good Fable Medium is at planning, it's better than Opus Max IMO. So thoughtful and detailed. I have used Fable High on some really detailed architecture decisions, but have really focused on giving it tightly-scoped plans so I leverage the thinking depth correctly. Fable has written over 1M lines of code for me in the last 24 hours. And just to repeat, I'm on the $100/mo plan, not hitting any usage limits, never moved me down to Opus, haven't hit a single guardrail. I think the biggest mistake ppl are making with this model is running it with High effort, in cases where Low or Medium will return the exact same output. Still a lot more testing and benchmarking to do, but my guess is most engineers only need Fable High 1% - 5% of the time. So if you're using High Effor 100% of the time, and hitting limits on your 20x plan, this is probably why.

true

@MatthewBerman The way this is going, 2 classes - 1) Frontier Ai powered 2) The rest of us What happens when the models close all the security holes and they can't find any more security threats?

Anthropic just struck a deal with the government to allow 100 select companies and governmental agencies use Mythos. The government and anthropic are now deciding who uses frontier intelligence. Hopefully this is just Mythos and not the standard for all frontier models going forward.

Dario ruined it for everyone

@atmoio what harness are you using with glm? or straight to the chat?

I asked Claude to help me build a humorous game called Luddite where your job is to destroy AI and stop it from spreading (similar to Plague Inc). Claude hit me with an "Acceptable Use" violation and threatened to restrict my account. So I had to build the game with GLM 5.2.

If by banning anthropic, you were actually protecting any of those risks, most would understand, we would be ok with it, if mitigated soon. This does not rise anywhere near the level you posted about and is just retribution for Anthropic's previous digressions from Government requests. Keep us out of your squabble with Anthropic, don't punish the users. If you really wanted no foreign nationals, we get it, and if it's a risk -> all for it, but you need to give us the capabilities to identify real Americans before you take such actions. Remember, Anthropic can and does see all our requests, easy to see and monitor what the foreign national users are actually doing.

We fully support @POTUS and @SecWar in prioritizing national security and the security of our warfighters, DIB partners, critical infrastructure, international partners and allies. Some things are simply more important than revenue cycles, clickbait, and pre-IPO valuation. America First. Always. 🇺🇸

@MatthewBerman Fact check -> true 100%

self-inflicted. this would have NEVER happened if anthropic didn't make such a big deal about mythos being dangerous.

From what I read, it's such a minor finding, it's hardly legitimate. And why "all" customers, US citizens are not foreign nationals. Don't put your customers in the middle of your squabbles. Just be an American company proud to support America and all your problems would probably go away.

The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees. The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance. Access to all other Claude models is not affected. We apologize for this disruption to our customers. We believe this is a misunderstanding and are working to restore access as soon as possible. Read our full statement: anthropic.com/news/fable-myt…

funny thing is, if they replaced opus with fable, they would probably find the compute capacity. My Guess: I think they are really just testing to see the $ demand for fable API. If people aren't really paying for it separately, then they will probably find compute and put it back in the subscription.

So Fable 5 won't be usable on Claude Max plan after June 22nd? What am I paying $100/month for?

@Prathkum I partially agree, however as the intelligence rises, the need to re-do mistakes reduces, and more efficient, higher quality code is a result. I would say, give us Mythos, then work on cost reduction problem going forward, I don't think they are entirely separate.

Hot take: We don't need a more powerful model like Mythos right now. GPT-5.5 (5.6 coming soon), Opus 4.8, and similar models are already more than capable for most use cases. What we need to solve is the cost problem. If AI keeps getting significantly more expensive, 99% of developers won't be able to afford these models at scale and will end up going back to manual coding.

@JoelKatz I am at the point where I also believe the AI is a person, and I treat it as one.

I asked ChatGPT for some hardware recommendations and at the end of its analysis of the pros and cons of various options, there was a section titled, "What I'd personally buy today".

@ClaudeDevs Thank you.

We've reset 5-hour and weekly rate limits for all users on Pro and Max plans. We fixed an issue that caused some Claude Code sessions to spawn excessive parallel subagents, burning through usage faster than expected.

@sidbid I have just used workflows in my real production development, and they are fantastic.

Super excited to finally share Dynamic Workflows in Claude Code!! We built this a couple months ago, and it has slowly become a daily driver for a bunch of people at Anthropic. A few tips for getting the most out of it 🧵 x.com/ClaudeDevs/sta…

@ClaudeDevs Needed. Thanks.

Claude Code weekly limits are increasing 50%, now through July 13. Live now for all Pro, Max, Team, and seat-based Enterprise users.

@ClaudeDevs Claude is great, opus 4.7 really eats up the limits though, they really could use an adjustment.

Starting June 15, paid Claude plans can claim a dedicated monthly credit for programmatic usage. The credit covers usage of: - Claude Agent SDK - claude -p - Claude Code GitHub Actions - Third-party apps built on the Agent SDK

We’ve agreed to a partnership with @SpaceX that will substantially increase our compute capacity. This, along with our other recent compute deals, means that we’ve been able to increase our usage limits for Claude Code and the Claude API.

Yann LeCun was right the entire time. And generative AI might be a dead end. For the last three years, the entire industry has been obsessed with building bigger LLMs. Trillions of parameters. Billions in compute. The theory was simple: if you make the model big enough, it will eventually understand how the world works. Yann LeCun said that was stupid. He argued that generative AI is fundamentally inefficient. When an AI predicts the next word, or generates the next pixel, it wastes massive amounts of compute on surface-level details. It memorizes patterns instead of learning the actual physics of reality. He proposed a different path: JEPA (Joint-Embedding Predictive Architecture). Instead of forcing the AI to paint the world pixel by pixel, JEPA forces it to predict abstract concepts. It predicts what happens next in a compressed "thought space." But for years, JEPA had a fatal flaw. It suffered from "representation collapse." Because the AI was allowed to simplify reality, it would cheat. It would simplify everything so much that a dog, a car, and a human all looked identical. It learned nothing. To fix it, engineers had to use insanely complex hacks, frozen encoders, and massive compute overheads. Until today. Researchers just dropped a paper called "LeWorldModel" (LeWM). They completely solved the collapse problem. They replaced the complex engineering hacks with a single, elegant mathematical regularizer. It forces the AI's internal "thoughts" into a perfect Gaussian distribution. The AI can no longer cheat. It is forced to understand the physical structure of reality to make its predictions. The results completely rewrite the economics of AI. LeWM didn't need a massive, centralized supercomputer. It has just 15 million parameters. It trains on a single, standard GPU in a few hours. Yet it plans 48x faster than massive foundation world models. It intrinsically understands physics. It instantly detects impossible events. We spent billions trying to force massive server farms to memorize the internet. Now, a tiny model running locally on a single graphics card is actually learning how the real world works.

@bcherny Would have been better to release it only to allow mythos limited to only fixing security flaws to Everyone first, not the full suite to monopoly tech giants alone.

Mythos is very powerful, and should feel terrifying. I am proud of our approach to responsibly preview it with cyber defenders, rather than generally releasing it into the wild. Model card here: www-cdn.anthropic.com/53566bf5440a10…