how do you know this is a better model? I wouldn't take any of the numbers at face value especially when all they have done is more/better post-training and thus the base pre-trained model capabilities is still the same. The model may just elicit some of the benchmark capabilities better. You really need to spend time using the model to come to any reliable conclusions.
the problem here is that text as the communication interface is not good for this. the model should be reasoning in the pose space (and generally in more geometric spaces), then interpolation and drawing is pretty easy. I think this will happen in some time.
Locomotion and manipulation are pretty different. The former we know how to do well -- this is what you see in unitree videos. Manipulation still not so much. This is not at all like GPT-2 because we still don't know what to scale (and even the data to scale is not there).
Here you can see another much simpler robot folding clothes for far longer: https://www.youtube.com/watch?v=gdeBIR0jVvU (there are more videos from other companies as well)
To answer your question -- folding clothes is easy, because clothes easily deform, do not break, fall smoothly when you drop them and most importantly are easily resettable task. Just through the well folded cloth up and voila start again.
Actually, folding clothes is a challenging dexterity task. However, it's a trivial mechanical engineering task, which is why it is so popular with underpowered robot arms.
How are you defining dextrous? I think it can be somewhat challenging but not dextrous -- the robot doesn't need to be very precise (few cms here and there do not matter), there are no forces involved, motions are all pick-place. Dextrous tasks would be things like shoe-lace tying, origami folding etc.
The current best neural networks only have around 60% success rates for small horizon tasks (think 10-20 seconds e.g. pick up apple). That is why there is so much cut-motions in this video. The future will be awesome but it will take time a lot of research still needs to happen (e.g. robust hands, tactile, how to even collect large scale data, RL).
Perhaps this is a bit pedantic, but what about the probable eventual proliferation of useful humanoid robots will make the future awesome? What does an awesome future look like compared to today, to you?
All with much improved privacy, reliability, order of magnitude lower cost, no risk of robbery/SA, etc. 24/7 operation even on holidays. Imagine service staff just sitting waiting for you to need them, always and everywhere.
Nevermind how much human lifespan will be freed from the tyranny of these mindless jobs.
In all of these posts there is someone claiming Claude is the best, then somebody else claiming they have tried a bunch of times and for them Gemini is the best while others find GPT-5 is supreme. Obviously, all of these are subjective narrow experiences. My conclusion is that all frontier models are both good and bad with no clear winner and making good evals is really hard.
* Gemini has the highest ceiling out of all of the models, but has consistently struggled with token-level accuracy. In other words, it's conceptual thinking it well beyond other models, but it sometimes makes stupid errors when talking. This makes it hard to reliably use for tool calling or structured output. Gemini is also very hard to steer, so when it's wrong, it's really hard to correct.
* Claude is extremely consistent and reliable. It's very, very good at the details - but will start to forget things if things get too complex. The good news is Claude is very steerable and will remember those details if you remind it.
* GPT-5 seems to be completely random for me. It's so inconsistent that it's extremely hard to use.
I tend to use Claude because I'm the most familiar with it and I'm confident that I can get good results out of it.
I’d say GPT-5 is the best in following and remembering instructions. After an initial plan it can easily continue with said plan for the next 30-60 minutes without human intervention, and come back with a complete working finished feature/product.
It’s honestly crazy how good it is, coming from Claude. I never thought I could already pass something a design doc and have it one-shot the entire thing with such level of accuracy. Even with Opus, I always need to either steer it, or fix the stuff it forgot by hand / have another phase afterwards to get it from 90% to 100%.
Yes the Codex TUI sucks but the model with high reasoning is an absolute beast, and convinced me to switch from Claude Max to ChatGPT Pro
Claude can do large code bases too, you just need to make it focus on parts that matter. Most of the coding tasks should not involve all parts of the code, right?
Personally I prefer Gemini because I still use AI via chat windows, and it can do a good ~90k tokens before it starts getting stupid. I'm yet to find an agent that's actually useful, and doesn't constantly fuck up everywhere while burning money.
Answer is a classic programming one - it depends? There are definitely differences in strength and weaknesses among them.
I run claude CLI as a primary and just ask it nicely to consult gemini cli (but not let it do any coding). It works surprisingly well. OpenAI just fell out of my view. Even cancelled ChatGPT subscription. Gemini is leaping forward and _feels like_ ChatGPT-5 is a regression.. I can't put my finger on it tbh.
In my experience gemini is good at writing specs it's hit or miss in reviewing code and it's not really usable for iterating on code. Codex is slow but can crack issues that Claude Code struggles with. So my workflow has being to use all three to iterate on specs. Have claude code work on implementation and have Codex review claude code's work (sometimes have gemini double check it).
Yeah, my take is it’s sort of up to the person using the LLM and maybe how they match to that LLM. That’s my hunch as to why we hear wildly different takes on these LLMs working for people. Gemini can be the most productive model for some while others find it entirely unworkable.
Not just personalities and preferences, but the purpose for which the AI is being used also affects the results. I primarily use AIs for complex troubleshooting along the lines of: "Here's a megabyte of logs, an IaC template, and a gibberish error code. What's the reason?" Right now, only Gemini Pro 2.5 has any chance of providing a useful output given those inputs, because its long-context attention is better than any other model's.
Capability wise, they seem close enough that I don’t bother re-evaluating them against each other all the time.
One advantage Gemini had (or still has, I’m not sure about the other providers) was its large context window combined with the ability to use PDF documents. It probably saved me weeks of work on an integration with a government system uploading hundreds of pages of documentation and immediately start asking questions, generating rules, and troubleshooting payloads that were leading to generic, computer-says-no errors.
No need to go trough RAG shenanigans and all of it within the free token allowance.
Because how good a model is is mostly just what the training data is at this point.
It's like the personality of a person. Employee A is better at talking to customers than Employee B, but Employee B is better at writing code than Employee A. Is one better than the other? Is one smarter than the other? Nope. Different training data.
so their way to differentiate against frontier labs is to try writing research blog posts (not papers). It will be interesting to see how this plays out. I don't think that anyone serious about developing frontier models would be putting anything useful out there for others. We already see this with all the incumbents -- Google, OAI, Anthropic, xAI, DeepSeek and other chinese labs.
This paper was just too overhyped by the authors. Also, the initial evals were very limited and very strange. This blog post does a much better job at a similar observation -- goes into details and does proper evaluation (also better attribution): https://jinjieni.notion.site/Diffusion-Language-Models-are-S...
and so is the safety margin for a humanoid. The consumer market is huge only if the robots are highly reliable and work very well both of which are not true at the moment. Things will change but it will take quite a bit of time and much more research.
Safety stops are much more challenging in a robot that tends to fall over without active balancing. Though with ISO 25785-1 in progress maybe there'll be a workable humanoid robot safety standard in a few years.
Interesting, hadn't heard of ISO 25785-1. It appears to be for industrial robots, though, and explicitly excludes "mobile robots intended for consumer or household use".
reply