Hans Gruber’s Parachute

Hello British Transport Police? I’ve seen something that doesn’t look right

Mythos removing the competition

Lo último que me esperaba de mi semana era darme cuenta de que mi profesor está metiendo prompt injection en los PDFs de las prácticas para pillar a los que las hacen full con IA… y que estaba cayendo TODO el mundo

👋 1h prompt cache is nuanced actually. It costs more for cache writes, and less for cache reads. Whether you benefit from cheaper cache reads depends on your usage pattern -- context window size, whether the query is the main agent or subagent, etc. We have been testing a number of heuristics to give subscribers better prompt cache hit rates, which means lower token usage and lower latency, when it works. But this effect is far from uniform due to the nuance above. Say you use 1h cache for an agent, but only used the agent to make a single query -- in this case the 1h cache would be wasted and you'd be overcharged. At this point we have rolled out 1h prompt cache by default in a number of places for subscribers to optimize cache duration based on real usage patterns, but we actually keep it at 5m for many queries also (eg. subagents, which are rarely resumed so you'd be paying for them even though they do not benefit from 1h). We also are not defaulting API customers to 1h yet -- this needs more testing to make sure it's a net improvement on average. Separately, when we do this kind of experimentation, we use experiment gates that are cached client-side. When you turn off telemetry we also disable experiment gates -- we do not call home when telemetry is off -- so Claude reads the default value, which is 5m. We will soon be changing the client side default to 1h for a few queries, since we now feel good that it is a small token savings on average for those queries. We will also give you env vars to force 1h and 5m. In any case, the token savings is nowhere near 12x unfortunately. It is a small win though, that we have been in the process of rolling out to everyone. Hope the explanation helps. More here: #pricing" target="_blank" rel="nofollow noopener">platform.claude.com/docs/en/build-…

The one employed philosophy PhD right now

okay I guess I have to talk about Péter Magyar here. Let me just start with saying, in a very unladylike way, that you guys seem to have zero clue what happened in Hungary in the last two years, you completely miss the point, and you're a disappointing bunch. Let's go.

I wonder if @anthropic @claudeai now asking this quite regularly is linked to the rollout of memory, having visibility of all the recent work that is added to memory persevered between sessions...

Thinking about this Bono quote again

@fleurmeston Where is this?

I was very restrained

@bzamayo or maybe there's a new one on the way...

The supply constraints are so bad that Apple currently does not let you order any desktop Mac with more than 96 GB of RAM. 9to5mac.com/2026/04/11/mac…

@waymo in London

anthropic: “we have finished training the ultimate god model exposing zero-day vulnerabilities in all software including linux and ffmpeg and we also made ten billion dollars last month” openai: “we have acquired TBPN”

@aakashxsh They did

Missed opportunity: Resident Eval github.com/milla-jovovich…

I did the math a couple weeks ago and it turns out a vegan prompting a frontier LLM *every second, 24/7* consumes less water than the average omnivore who never uses AI.

Sex is good but have you tried saving California’s only nuclear plant and sticking it to the boomer anti-nuclear environmentalists?

Was looking for buildings inspired by Escher and came across this Chongqing bookstore.

M&S "Food" 🫠 😭

This is wild. Google Research demonstrates a ~20x more efficient implementation of Shor's algorithm that could break ECDSA keys within minutes with ~500K physical qubits. Google is now are more confident on a 2029 post-quantum transition. We are no longer looking at mid 2030s, we could have quantum computers of this scale by the end of the decade. They believe this result is so severe that they are not publishing the actual circuits. They instead published a ZKP proving that they know of the quantum circuit with these properties. This is very atypical, showing Google thinks this is serious shit. All blockchains need a transition plan ASAP. Post-quantum is no longer a drill.