And by benchmarks (unless they gamed them), seems to be at around Opus 4.7 level, which is what Elon mentioned in https://x.com/elonmusk/status/2074911038286295049.

I guess the Cursor data was very useful.

> Training included trillions of tokens of Cursor data which capture a wide-range of user interactions with codebases and software tools. This dataset lets the model learn both from existing software as well as developer-agent interactions, capturing how developers work and how agents interact with their environments.

This is what the big money was for. Cursor is the first big player that had real-world data from real-world projects, before cc / codex were a thing.

> We used reinforcement learning on difficult problems in realistic environments spanning both software engineering and broader knowledge work. These environments teach the model to investigate problems, use tools, recover from mistakes, and verify results.

> Many of these problems had to be designed to be difficult enough that even frontier models fail at them. As models improve, existing tasks stop teaching them anything new, and problems that once required extensive reasoning become routine.

> We developed a distributed agent system to construct these environments at scale. Engineers specify a problem and how a solution is verified, and large groups of agents construct, test, and refine each environment.

This is where scale comes in. You use the previous gen model to prepare datasets for the next model iteration. The better the models, the better the data, the better the next models. (they also have a comparison with their composer2.5 training run, for people still thinking chinese models are "close to SotA"...)

Reports of xAIs demise (after giving a lot of compute to Anthropic) were slightly exaggerated, it seems.

> Grok 4.5 was trained across tens of thousands of NVIDIA GB300 GPUs

- Very fast, easily beats GPT 5.5/Opus 4.8/GLM 5.2 because of higher t/s (around 90?) and very high token efficiency

- Very good price, no contest vs GPT and Opus which are very overpriced if you pay API costs, and probably cheaper than GLM 5.2 when you take into account the token efficiency.

- Will take quite a while to get a feel for how smart it is, but it's definitely good, I'd say in the same tier as opus, occupying the lower end of that tier together with GLM 5.2.

(I am not an iOS developer, so getting something specific that I needed in a few hours/days was really helpful instead of spending months/years learning the language, APIs, etc.) (I am absolutely not "vibe-coding" Caddy btw, just tinkering with it for personal projects.)

Notably:

> Grok 4.5 and Composer 2.5 are two different model weight classes, and we're excited to support both sizes and weights. Composer 2.5 will remain offered, and we will release new models of this size going forward.

Edit: Gemini 3.5 Pro. Expectations grow with each day it is not released.

This -- training on work done on hard, real-world tasks -- seems to be how most frontier models are making capability gains these days. In fact people make decent money doing that for data companies like Mercor. However it's also striking that Cursor managed to gather so much of such data.

Turns out Cursor will train on everything you do unless you opt-out, even if you're already paying for it with cash! Are that many people really not opting out?

This is why it seems like a significant concern to me: It's very clear that typical, run-of-the-mill coding has been completely commoditized, so the primary value remaining is either in novel use-cases and applications, or novel technical solutions to hard problems.

Presumably the value for novel use-cases could be captured by building a business around it via the usual moats (distribution, relationships, network effects, first mover advantage, etc.) so the code and techniques do not matter as much.

However novel technical solutions, which are already hard to monetize without building a whole damn business around it, could at least be capitalized on by simply being able to claim credit for it. I'd at least like the option of being "paid in exposure" if I'm not getting paid in cash. But having them "leaked" unwittingly via the training corpus to whosoever happens to prompt the model with the same problem removes even that option.

I know people have been calling out this risk forever, and I don't use any tool that I can't opt-out of training completely, but the scale at which this is happening -- on an ongoing basis, mind you, after training on the data of the whole world, and that too after paying for the product -- is surprising. I'm bullish on the technology but we really should be way more careful handing these AI companies even more of our intellectual crown jewels.

This is a model I could really see used inside applications, where Opus or Sonnet or GPT-5.5 are too expensive.

I would really like to see a strong Deepseek v4-Flash competitor, which ideally is something like Sonnet 4.6 performance at <$0.30 per token. This is missing from main US labs.

The vast majority of input tokens are normally cached, so this is actually the price that matters. I wonder why it's so high.

GLM 5.2 caught up, Cognition RL'ed Kimi 2.7, Grok 4.5 is out, DeepSeek v4 GA is out in a few days...

What is the moat? and why should we pay for the expensive tokens today instead of just waiting a few months/weeks and getting AI for significantly cheaper?

I must say, I feel like companies spending Millions on Anthropic tokens are just negative capex'ing and wasting money, even OpenAI is barely ok pricing...

I'll give this one a try with a grain of salt and lowering my levels of expectations

Not enough people are noticing this, they juiced the benches

[1]: However it does say "Requires user IDs" under anonymity, which is unusual on OpenRouter and not something I particularly like to see. Generally, OpenRouter is a proxy that anonymizes requests to providers, and I can't find an account-wide setting to enforce that like ZDR-only.

Most labs - including OpenAI and Anthropic, but also Google and Chinese labs - highlight their scores in benchmarks that have fixed, widely available answers. Those answers end up in the training data and so models can just regurgitate training data instead of actually doing the benchmark. As a result, most benchmarks often quoted are essentially meaningless for gauging model performance.

Terminal-Bench still publishes answers, but neither DeepSWE and SWE-Bench Pro do. Especially for DeepSWE it's been difficult for models to fake good results so far. SWE-Bench Pro does have weird outliers like good performance for e.g. the atrocious Muse Spark, but it also doesn't provide answers for the training data.

So either they're good, or they found a way to game DeepSWE. Given that the Cursor team previously published the well-received Composer 2.5 a good score here doesn't come out of nowhere, so this might hold up. Cursor has enormous amounts of training data to train good coding models with.

[0] https://docs.x.ai/developers/models/grok-4.5

[1] https://www.anthropic.com/news/claude-opus-4-8

Grok: "I can answer any math question in less thant 500ms" User: "what is 2319321x232" Grok: "3201521321" User: "this is wrong" Grok: "but it was less than 500ms""

I don't really understand what Grok Build is? Is it an API? A CLI? What?

Just really dissapointing hysteria on display here.

Like it or not, Elon and his companies have changed the world for the better.

You are allowed to separate the human from the human's impact sometimes.

EDIT: After looking at my own usage stats - I stand corrected! It is under the "Auto + Composer" tier - brilliant!

I think we are going to be waiting a long time for Twitter / X to go bankrupt as it was (erroneously) predicted a long time ago.

Did anthropic found their moat or we hit a Wall?

terminal is nice but codex desktop app is very useful

The amount of rule-breaking comments in this thread is pretty much out of control.

Where have the hardcore nerds gone? How is the model, is it good at coding? What does this mean for competition and pricing?

Why give them money?

It would be one thing if they were the only game in town but thats definitely not the case.

Google Deepmind has failed.

Flash 3.5 seems capable for a flash model, Antigravity seems like a reasonable harness. But GDM is responsible for the frontier model and it looks like a complete failure.

What's particularly galling is the size of funding of GDM. It is enormous compared to the other labs. The headcount of other labs is swollen by infra, marketing, sales, GDM is pure "engineering" and its frontier model isn't even leading open source.

What a failure. It's unreal.

I asked it today to fix a non-simple bug and MAI fixed it in one shot with less than 70k tokens (Cursor would have used probably half a million tokens based on my previous usage). Orgs need to start getting more visibility into why Cursor burns so many tokens.

It's an excellent model. GPT 5.4/5.5 level, some things better, others not, but extremely fast. A wonderful technical improvement.

If a Chinese company or random startup released the model, people would be glazing it like crazy.

xAI is competently keeping up with the frontier, just as well as any of the Chinese labs or Mistral. Given any significant breakthroughs, xAI will be better positioned to capitalize on them than nearly any other entity.

I can't wait to see what Meta comes up with; with 4 contenders in the US race, we'd have a lot of be grateful for.

Their inital image generation was a wrapper around Flux.

People don't buy it any longer, just like no one bought the fake SpaceX stock recommendations yesterday and everyone just sold.