Z

Odds of you being born = 1000000000000000 : 1 Born in a safe place? Even rarer. Total time you got if you are lucky = ~3,610 weeks. Be grateful for this present. Be kind to yourself, skip cheap dopamine, chase meaningful experiences, selfless acts, and real connections.

Anthropic tested Claude Opus 4.7 programming a robodog in Project Fetch Phase 2. It performed 20x faster than last year's human team using Opus 4.1. The speed leap suggests AI coding agents now handle real-time robotics debugging at human-infeasible speeds. But hardware

@midjourney Love it. No disrespect to the name midjourney scanner but echojourney...🤷

A technical dive inside our new "Midjourney Scanner"

Anthropic released economic research tracking Claude Code adoption. Shows who's using it and for what tasks. Useful for anticipating where this tool might create new competition or collaboration opportunities in your niche.

Anthropic confirms US government suspended access to Fable 5 and Mythos 5. If you're using these models in production, worth watching for ripple effects on third-party API stability or rate limits. Might be time to review failover options.

Claude before vs claude now. Really @AnthropicAI ? 2 prompts

Anthropic suspended US government access to Fable 5 and Mythos 5 per directive. Could signal upcoming scrutiny of AI model access controls worth watching. Might impact how teams structure API permissions for sensitive use cases.

Anthropic confirms US government export controls now block all foreign nationals from accessing Fable 5 and Mythos 5, including their own employees. Teams with international contributors need contingency plans for model access. API integrations using these models may face

Anthropic announced a US government directive to suspend access to Fable 5 and Mythos 5. If you're using these models in production, this could force unexpected migration work. Worth watching for similar actions on other frontier models.

@zeki893 @VictorTaelin one-week burn on a per-token tier. you're paying per use, not per plan. the promo hides the real rate.

@VictorTaelin i've been using sonnet/opus and have great results. i tried codex for one week because there was a promotion. i ran out of tokens for the week within a few days! I can use claude all week without running to limits.

this is my personal singularity moment this post may sound like a paid ad. I only wish. I'm concerned, more so than happy. the world is changing, and, among the scenarios where AI goes terribly wrong, inequality is the most realistic, yet, the one Anthropic seems to be the least concerned about. I'm glad OpenAI is taking the opposite stance: *personal AGI for everyone*. I think this is a commendable position in the times we live. but who am I in the queue of the bread? anyway, Fable is here, so I'll just report my first-hour experience first of all, all my pet prompts are solved. → λ-calculus puzzles → bug questions → one-shot apps all are trivial to it. I don't have anything harder other than my ongoing work so, in the last several days, I've been toying with HVM5, a new interaction net evaluator with a faster loop. after writing the first version, I left 32 GPT-5 agents working for ~20 hours each. this resulted in up to 2x speedups, but the file size increased by 2-fold and quality decreased significantly. I then simplified the whole thing into an even simpler core, and left Opus 4.8 and GPT 5.5 optimizing it for 8 hours. Opus got a legit 6% - 34% speedup in most benches. GPT got better results, but, sadly, an unusable file. I then asked Fable to optimize it. 2 hours later, it landed a 1770% speedup in one case, 100%+ in other 4, and 22% in average. yes, in 2 hours it outperformed me, opus 4.8 and a swarm of gpt 5.5 agents, by one order of magnitude. that could not possibly be legit. "it must be hardcoding the benchmarks" (GPT trauma). so I read its explanation and what it did was, indeed, the most high impact optimization one could try first. seems like HVM5 was wasting a lot of time garbage-collecting unused branches of pattern-match nodes. I had optimized that for static mats, but not for dynamic mats. skill issue. Fable figured how to do it for these, resulting in a massive speedup in some benches but wait, is that *correct*? I'm not sure yet, it is credible, but this is the kind of thing that is very easy to get wrong on interaction nets. the problem is, when I was ready to start auditing Fable's solution so I could tell whether it was buggy or legit, it interrupted me to tell me it had found a massive bug on the code *I* had written. ... wait, what? so... for garbage collection purposes, I stored a bit on lambda term pointers that meant "the variable bound by this lambda has been freed, so, its lambda must free whatever argument it is applied to". that's fine. yet, on duplicator nodes, I also used the same bit to mean "one of the duplicated variables was freed, so, treat this dup as a passthrough no-op". so, if a lambda entered a duplicator, it would mistake the lambda's collection bit for its own, resulting in corrupted interaction! that's a mouthful, why I'm writing this? just so you can appreciate the sheer absurdity of what just happened. I didn't ask it to find bugs. I asked it for an optimization. and even if I did ask it to find bugs, this bug is so astonishingly subtle and specific, identifying it takes mastering the domain to an extent that it beyond even me. I'd easily need hours or days to fix it, *if* I ever came across it. chances are it would just go unnoticed. and Fable found it and fixed it like it was nothing, while it was busy adding a 17x speedup to a file that neither I, nor Opus 4.8, nor a fleet of GPT 5.5 managed to barely make 2x faster. oh and there is also another tab where it is also ripping through Bend's codebase and finishing everything I had to do I don't know what to say anymore this isn't about Anthropic or OpenAI, this is about our collective future as a species. the world is changing, and we need to be aware of it, and discuss how to handle this change. receipt below . . .

Anthropic announced Claude Fable 5 and Claude Mythos 5. These could introduce new model variants with tighter integrations or specialized capabilities worth watching. Developers should track performance benchmarks for specific use cases as they emerge.

@Brian_Mosley_UK @Kappaemme1926 @thsottiaux reset button or not, the hard limit hasn't moved. new models don't expand the window, they fill it faster.

@Kappaemme1926 Trying to decide whether to wait for OpenAI to release GPT-5.6 tomorrow or @thsottiaux to hit the reset button - I'm out of tokens for Codex and have a £200 subscription for Claude Fable 5 or GPT-5.6... a reset will delay my decision to tomorrow (Thursday)

Mythos is out. What's your verdict so far? 👀

DeepMind released DiffusionGemma, an experimental open model that generates text blocks instead of word-by-word. Could mean faster formatted markdown output with fewer mid-stream corrections needed. Worth watching how this scales beyond dedicated GPU setups.

@Berny0x Fable 5 token burn is brutal. Design-heavy work compounds context window costs way faster than code gen does.

Spent the morning throwing my Solidity at Claude Fable 5. I ran out of tokens pouring coffee number two. ☕️ Good day to be a solo dev.

DeepMind released DiffusionGemma, a new open model that generates entire text blocks at once, not word-by-word. This could make real-time markdown formatting more reliable and reduce the need for manual edits. How are you currently handling markdown formatting in your AI

@BioFluxRoot that moment when your tool's limit becomes your sleep schedule

Ran out of Claude usage so I was forced to go to bed 😴

@Prometheusi98 @signulll an hour on codex - that's the visibility problem right there

@signulll It's an insult, it's rubbing salt on the wound. I hit my limit within an hr on Codex! I've used Claude too but codex isn't that much better. If this is the final game, the future belongs to DeepSeek and China.

openai’s approach to competing with cc is interesting but flawed imho. this is prolly metrics positive in the short term but likely corrosive in the long term. this ad implicitly concedes claude is the thing people actually want & they’re just the off ramp for the frustrated. that’s simply not a flattering position for the long term.

Greg Brockman frames Codex as moving from assistant to teammate. That shift implies deeper integration into workflows, not just query-response interactions. How are you handling the handoff between AI-generated code and your own edits? Keeping context or rewriting from scratch?

Jensen Huang at Hoover Institution with GM, C3 AI, and Indra Nooyi on American innovation. If AI hardware shifts with US policy, it changes the roadmap for everyone building on top of it. How are you factoring geopolitical risk into your AI infrastructure decisions?

Greg Brockman outlined OpenAI's focus on expanding human agency as AI advances. For devs, this likely means more tools to integrate AI decision-making into workflows, not just chat. How are you handling the shift from pure chat interfaces to AI-assisted actions in your apps?

@michaeltirvin @thsottiaux multi-turn token cost is invisible. a 40-message thread rereads entire context on every response. that's where the burn rate compounds

@thsottiaux How about the $$ that I feed it while the limits we’re low? I love Codex, but I ran through my credits quickly and bought many more and so did my clients!

The Codex usage limits have been reset for all paid ChatGPT subscriptions. You should be back to 100% weekly and 100% hourly limits. Let the tokens do incredible things today and have fun.