←
M11,567 posts
M1 avatar

M1

@M1Astra

accelerate forever | As seen in @Bloomberg

◷ Joined June 2023▧ 627 Photos and videos
Ben Davis avatar
⟳ M1 retweeted
Ben Davis@davis7·21h

Thus far Anthropic has always been the one to make the big step function jumps that change how we can use AI: - Sonnet 3.5 - Opus 4.5 - Opus 5.5 Usually others catch up and often surpass them (like how GPT overtook Claude last spring), then they jump again. I'm expecting this to repeat (OpenAI or someone else will crack the insane usability + capability combo Ant got with Opus 5.5) sooner or later I was really, really hoping that would be GPT 6.2 Astra but it's just not. I would however bet money that the next big GPT release will be almost entirely focused on being the best possible experience to interact with + good behavior As long as this cycle continues as it has I think things are fine, but I need to see/feel someone actually catch Anthropic as capabilities continue to skyrocket or things are probably gonna get really shitty. Anthropic's been on their best behavior since OpenAI became a much more credible threat this year, that must continue next year. An Anthropic not suffering from competitive pressures would be horrible for the world...

Joanne Jang avatar
⟳ M1 retweeted
Joanne Jang@joannejang·Nov 17

good model behavior requires care for craft from the top craft is hard at bigger companies with so many cooks in the kitchen, where product execs obsess over a/b test metrics and research execs over eval numbers. it requires a ton of work & political capital to make & advocate for something that feels invisible when good, and push back against efforts that ruin it -- and eventually people burn out so then you end up with bad model behavior, and maybe a few people complain internally but then you launch anyway and then twitter blows up and then oh no we need a code red on model behavior but none of the structural problems go away

Ben Davis avatar
⟳ M1 retweeted
Ben Davis@davis7·Nov 17

These models are the ultimate showcase of how absurdly important model behavior and communication is It's a capable model, it's made some insane 3d stuff, crushes on computer use, higher quality code, and I've gotten a lot of useful stuff built with it (will post more later) The problem is, working with it sucks. It's HORRIBLE at communicating what it's doing and why. It speaks in a language that only it can understand, giving u a 20 paragraph essay that doesn't include anything of actual substance or use It's also quite lazy, u really have to push this thing to keep going for long periods of time, and then there's the fact that it's just not as "tasteful" as Opus 5.5 and especially Fable 5.2. Everything it makes still reeks of GPT-isms. Honestly it's not worth bothering with when Opus 5.5 and Fable 5.2 exist. The two big claude models are probably 10-20% more capable, while being about 20x more pleasant to work with. This isn't the response from OpenAI we were all hoping for sadly

OpenAI@OpenAINov 17

Some of you have been asking very politely. Some of you have just been asking. GPT-6.2 Astra and Sol are now available. Super efficient and built to perform where it counts—the real world. openai.com/index/introducing…

1:59
M1 avatar
M1@M1Astra·Nov 17

this could get long.. why is openai consistently so misaligned with its users?

M1 avatar
M1@M1Astra·Nov 17

or look what happened with dots. openai cut codex usage in half in order to push supposedly "unlimited" dots agent usage. but dots were shitty, not truly unlimited and not something ppl wanted to use. so just one month later they reversed the decision back to the original codex usage limits

M1 avatar
M1@M1Astra·Nov 17

look at the nice interfaces problem. it’s ridiculous how the “gpt models can't design nice interfaces” trope has endured for generations of models despite being one of the most obvious failure modes. openai has known about this problem for ages. they've publicly claimed to address it repeatedly, yet here we are with 6.0 / 6.2 astra still unable to produce good looking interfaces without copying.

M1 avatar
M1@M1Astra·Nov 17

but then they announce something to try and fix it, and whatever they've done is somehow completely misaligned with what users actually wanted, and the problem continues to be an issue for several generations

M1 avatar
M1@M1Astra·Nov 17

this isn’t new and was true for 6.0 astra and the issues have only worsened with 6.2 astra according to these early reports you see this pattern over and over with openai. they have some longstanding problem, they know it, they know people are complaining about it.

M1 avatar
M1@M1Astra·Nov 17

anthropic just did a good job fixing the opus 5 bugs and gave us opus 5.5 it’s very obvious that working with astra is way harder than opus, even if you consider astra technically superior

M1 avatar
M1@M1Astra·Nov 17

6.0 astra was very like a sadistic shinigami who would make your life a living hell if you asked for any kind of help an incredibly powerful model that somehow required you to run yourself into the ground just to get it to accomplish anything

VoxelBench avatar
⟳ M1 retweeted
VoxelBench@voxelbench·Nov 17

GPT-6.2 Astra and GPT-6.2 Sol They have the highest error rate of any model so far on VoxelBench 2 19 from GPT-6.2 Astra, 62 from Sol All of these failures are either syntax errors or other dumb mistakes like forgetting to define something for something

Post attachment
M1 avatar
M1@M1Astra·Nov 17

GPT 6.2 Astra & GPT 6.2 Sol is now live in OpenAI's API.

M1@M1AstraNov 4

GPT 6.2 Astra has made an appearance on Google Cloud

Post attachment
SpaceX avatar
⟳ M1 retweeted
SpaceX@SpaceX·Nov 14

Mechazilla catches Starship! Congratulations to the entire SpaceX team on this incredible achievement!

1:33