reckless

5.6-sol would be a better comparison given it's general availability and usage allowances

Takeaways Which model to use for generating architecture.

The simpler the decision you need to make, the more readily you can just take a proposal from Fable or GPT-5.5. And now that Fable is unavailable outside the US, the choice between Opus and GPT is far from obvious. As a quick default I'd lean toward GPT.

Doesn't Claude have memories like Codex?

No way is Muse Spark generally better than offerings from Google and OpenAI. I actually find arena to be amongst the most useless indicators

Codex has been great for me

It's lawful use with specific laws called out though? New laws won't supercede what is agreed in the contract at the time of signing.

Anthropic wants to enforce them via language of the contracts and take a hands off approach. OpenAI has a contract that is paired with humans in the room (FDEs) that can pull the plug.

For people looking for a more vanilla experience I'm publishing Google's official extension supporting Chromium builds for Android here https://github.com/andrewginns/chromium-browser-snapshots-An...

Combined Obtainium it's easy to keep it updated. https://github.com/ImranR98/Obtainium

The aggregate picture only tells you so much.

Sites like simonwillison.net/2025/jul/ and channels like https://www.youtube.com/@aiexplained-official also cover new model releases pretty quickly for some "out of the box thinking/reasoning" evaluations.

For me and my usage I can really only tell if I start using the new model for tasks I actually use them for.

My personal benchmark andrew.ginns.uk/merbench has full code and data on GitHub if you want a staring point!

Been a long time user of Chatbox as my Android client of choice. The recent web search update has been fantastic!

Looks similar to what I've been using for a few weeks https://github.com/miurla/morphic

Does this also count platform costs or just chip cost? I'd imagine the threadripper motherboard and ram costs aren't insignificant

They do a 55" curved one called the Samsung odyssey ark

Indeed they do, however companies like Meta (altruistically or not) are preventing OpenAI from building 'moats' by releasing models and architecture details in a very public way.

True but what you can do is SSH to the device and install a custom launcher for apps that can read standard epubs, play chess, or expose the linux terminal on device.

Not great for basic users but I've had significantly more use out of it with some advanced setup.

Take a look at Piper. It's the tts solution used by the open source home automation project HomeAssistant. Produces decent quality speech in a couple seconds on raspberry pi class hardware.

It's fascinating how they managed to reclaim so much die area by dropping the cache and optimising for low (~3GHz) frequency.

I'd recommend Sweethome3D to trace the floorplan and then you can export it as a .obj to a program like Twinmotion to do visualisation and lighting with pathtracing etc.

Seems to be entirely a different approach for diffusion.

DeepFloyd IF works in pixel space. The diffusion is implemented on a pixel level, unlike latent diffusion models (like Stable Diffusion), where latent representations are used.

I'd definitely recommend you use homeassistant as a front-end for this. You can then update a 'sensor' using an mqtt message.

I feel like the comments on backwards compatibility are due to the absolute shitshow of TF2 compatibility for TF1 code and models.

Also the threat of Pytorch can be seen when reading between the lines, especially since it's now run by a foundation and the darling of the diffusion model developments.

Would be great to be able to utilise outpainting to generate larger images in smaller tiles at full precision.

I wonder if we see more of an adoption now that Zen 4 is confirmed to have AVX-512 support.

TL;DR is that AVX-512 provides >20% speedup compared to AVX2

It's a shame that intel isn't including AVX-512 on the 13th gen Raptor Lake consumer CPUs, and the newer skews of 12th gen like the KS processors have it fused off.

Are you also planning on open sourcing thr algorithms you're using to process the data? I'd be interested in helping out on the ML side for any sensor fusion or image classification.