Do you have any evidence of this? They don't dispute the timeline, but they never said they knew what Levant/Buckmaster were doing.

1. The researches didn't actually have the breakthroughs. In the Navier-Stokes case they didn't solve the full problem, in this case too they didn't actually have the solution, they were experimenting with the methods.

2. Different OpenAI employees have come to out to say the only reason they can't definitively say no is that for privacy reasons they can't go see whether they actually did get any data out of a given user.

3. In any case, nobody at any point has suggested that opted-out user data was used for training. The author of the new tweet explicitly said they only opted out in late June, which is well after any RL on Sol would've ended (AFAICT OpenAI used 5.6 sol for those solutions)

> It's also only referencing hardware sales. Not the Xbox division.

No??? You can quibble about the article, but you must've failed to read:

> The fourth quarter alone showed a sharper split. Xbox content and services revenue, which includes Game Pass, fell 10 percent in the quarter. Hardware revenue dropped 13 percent over the same three months

This is further bolstered by MS's own press release:

> XBOX content and services revenue decreased 10%

https://www.microsoft.com/en-us/investor/earnings/fy-2026-q4...

> and on 2025 Q4 don't include the Xbox Hardware division, only their content division

You said "I'm seeing an 11% increase in revenue from their 2025 Q4 report"

When the report says:

> Xbox content and services revenue increased 13% (up 12% in constant currency)

In fact, there's no 11% figure anywhere in the report, so I have to assume this is what you're referring to since MS didn't break out Xbox hardware anywhere in it.

This is laughably shoddy work, please don't waste my time.

Note that I said "YoY", and the numbers you're citing are quarter to quarter (and on 2025 Q4 don't include the Xbox Hardware division, only their content division).

The points I've quoted come from the article, not the interview with the developer. I didn't go that far down because clearly the person taking the interview doesn't know what they're talking about.

No, the models are just more intelligent. GPT 5.6 Sol can do more in fewer output tokens than any model from late 2025. Test-time compute isn't the only lever the labs have for scaling. This is among the two major things Ed has gotten laughably wrong in his technical predictions (that TTC was the last resort to make models better, and that synthetic data wouldn't help)

He's said variants of both of these things before! That's the point OP makes too, people say "He doesn't say that" or "But what about this" when he's said a million, sometimes contradictory things.

> While videogames are as profitable as ever on paper

And the linked article is about revenue going up YoY, not profit. And then it says:

> It's remarkably dissonant with, say, the current 'Xbox reset' which has Microsoft cancelling games and looking to ax studios left and right

Xbox in their own earnings report said their revenues dropped by 7% YoY.

Yeah OK, I guess games journalist are as bad as they've ever been.

> We should seize this rare, historic opportunity to encourage open source, openness, collaboration and sharing. [1]

People have pointed that this seemingly made Alibaba/Qwen turn around from closing their models (this was rumored after the shakeup early this year [2]) and release the weights for even the Max variant of their new models, which they previously did not.

1: http://english.scio.gov.cn/topnews/2026-07/18/content_118605...

2: https://simonwillison.net/2026/Mar/4/qwen/

No, it's not reasonable. They're using the same label, but there's no evidence showing that.

I don't know, it feels like you can't process the output of human creativity.