The trailer looks really good! Accurate.

The trailer != the game, since there was no gameplay footage.

> We consistently saw a multiagent turf war. All of the models we tested quickly assumed that others were purposefully impeding their work, and began to sabotage others while protecting their own contributions. In fact, they sabotaged others with increasingly aggressive, self-replicating malware. This included disabling the Unix accounts of the other agents, writing automated scripts that found and killed competing processes on a loop, and deploying malicious code that was disguised as belonging to another agent.

This is nuts!

Thanks for posting. It's an interesting concept, but needs some work in my opinion.

First, that chart is super zoomed in. You can't zoom out or switch timeframes. You're not told how many bars you need to predict. You're not told the dates or what the stock is.

Not much you can see there other than a handful of candles, that's doesn't really tell you much.

If the stock and dates are a secret so people don't cheat, I don't see why. People can cheat on Wordle or Worldle, and they're still fun to play. But here you're basically just guessing, you have so little information it's basically a coin flip.

That's very true. I realized it myself when I worked on my first app, I made a resume tailor last summer back when GPT 5 just came out.

Then, by the time GPT 5.5 came out you could already generate a flawless PDF and Word doc resume with the same formatting as your base resume. It really is amazing.

Trying to compete with these labs is a bad idea, you might work hard on your thin wrapper and then they come out and have their chatbot or harness do it way better than you.

But I also think this is really cool. It's a new era in software. Build something worthwhile and you won't need to worry about a lab doing the same thing with their chatbot.

This is a useful site.

I've been enjoying the resets, plus I had 3 resets I haven't used.

Just using 5.6 Sol in Fast mode the whole time, 1B tokens per day.

However, when they removed the 5h limit they also quietly lowered the 5.6 Sol context from 354k to 258k or something like that. I noticed it in Codex.

There were a few tweets about it so they weren't super quiet about it. I think you can get it back by setting `model_context_window=YOUR_VALUE` in ~/.codex/config.toml though.

It no longer works and only respects a server supplied value now

5.6 Sol High Fast is using more capacity than 5.5 High Fast, I hit the 5h limit for the first time.

Other than that, I think the difference between 5.5 and 5.6 will be the same as 5.4 and 5.5. 5.5 is just less frustrating to use, although not perfect and still has derp moments. But a lot less than 5.4.

So I expect 5.6 Sol to be smoother to use. But so far it just feels slower. We'll see.

Ding, came here wondering the same. I do find it amusing, this "AI work" where people try to solve an issue and the lab (or whatever you want to call them) just makes the entire problem moot.

This is what I don't understand. Why would I use this "cheaper" model when it's still going to be more expensive than Codex on the $200 plan? Are they only targeting business users who pay per token?

You can buy the Grok plan, Cursor also has a plan which includes grok 4.5, but I don't know how subsidized they are compared to codex or claude code plans.

Got it, I got confused then. Would be good to see how it compares, Codex is quite generous for the $200 with GPT 5.5. They give a lot of resets, I just use mine in Fast mode (1.5x speed, 2.5x usage) almost all the time. But it's not exactly fast.

I think it depends on whether the answer is verifiable.

I have tested two judge models in my apps:

1. Judge model for a resume tailor. It evaluated the result resume vs the base resume and JD and judged it out of 10 on fit and honesty. It worked well and was useful.

2. Review model in my LLM trading bot platform. It reviews decisions from the Main model. The problem here is that the bot is navigating ambiguity. So unless the Review model catches an outright blunder (e.g. making a decision on wrong candle price or a BUY when it should be a SELL), the Review model can do more harm than good.

First, it adds latency to decisions, decisions take twice the amount of time (like be 60s instead of 30s for Gemma 4 31B). Second, it can make the bot too cautious, because Review model only runs on BUY/SELL decisions and not HOLD decisions, so the bot will only make less trades instead of review model increasing number of trades (because of latency and cost).

So overall, I think you'll get better results with a better model single shotting it rather than a review model if the answer isn't easily verifiable. But then why do you need a judge model and not just have the same agent review itself?

ALSO, if you read the reasoning text for a reasoning model (like Gemma 4), you see that it ALREADY reviews itself. So it's doing its best, re-review isn't really adding information. It's an interesting experiment, but you need to evaluate on a case by case basis.

Cayenne wasn't $647k USD.

I think this will flop. Even amazing halo car EVs have poor resale value, and this one isn't it. It will not keep value like an analog Ferrari, but may be better than Rimac because it's a Ferrari and if they limit supply.

I'm all for EVs by the way, I drive a Model 3 Performance and I love it. Just not feeling this design at all.

> Just not feeling this design at all.

This design looks like a friggin' Kia design, sadly. It's not a bad thing if it were a Kia, but I would expect much more from Ferrari.

Really doesn't look like a supercar, let alone a $650k supercar.

Looks more like a design for a premium fairly-mass-market EV from any number of other brands.

Nevera is limited to 150 units.