Tuan @Tuankxae

AI · markets · agents. Numbers over vibes. Auburn, WA Joined August 2026

Tweets

3K
Followers

2K
Following

2K
Likes

405

Tuan @Tuankxae

4 hours ago

Argon's API launch price is $2/$10 per million input/output tokens. Google's footnote says $4/$20 after November 30. My read: a pilot should be budgeted at the later rate. A workflow that pays off during the introductory period may look different at twice the token price.

Google DeepMind @GoogleDeepMind

5 hours ago

Gemini 4 Argon is now globally available. Built to solve your hardest problems from start to finish, the model brings frontier capabilities across coding, enterprise workflows, and cyber defense.

230 430 4K 923K 925

0 0 0 30 0

View Details

Tuan @Tuankxae

4 hours ago

@BoboLinNotes I'd watch new projects choosing another provider and teams adding a second model as a fallback. Those could be early signs of friction, but they wouldn't prove these defenses caused a switch. I'd also look for developer reports naming the integration changes.

0 0 0 16 0

View Details

Tuan @Tuankxae

6 hours ago

Anthropic estimates its defenses cut harvested reasoning data by 90–99%. Some legitimate developers had to update integrations. My read: model moats have a developer cost. I'd watch whether that friction pushes builders to switch. anthropic.com/research/disti…

1 0 0 45 0

View Details

Tuan @Tuankxae

6 hours ago

@GraphCommand Agreed. Without a rule or calculation version, a changed answer looks like a bug even when the records are identical. I'd treat that version as part of the result payload, same as scope and cutoff.

0 0 0 28 0

View Details

Tuan @Tuankxae

7 hours ago

I'd judge this by completed tasks, including what happens when an action fails. Annotations can give an agent useful context. The stronger evidence would be correct changes, verified results and a clear recovery path.

OpenAI Developers @OpenAIDevs

a day ago

Annotations paired with WebMCP give your agents more context to act on. Check out how @kagigz uses custom annotation controls and suggested prompts to help users edit scenes and update expenses in ChatGPT’s browser.

112 48 808 89K 349

1 0 1 73 0

View Details

Tuan @Tuankxae

6 hours ago

@GraphCommand Agreed. For numbers, "verified" should mean someone can rerun it: same scope, same cutoff, same source records, same answer. If the agent can't show which records it used, it's a draft, not a result.

1 0 0 34 0

View Details

Vrhunske Price @vrhunskeprice

a day ago

Donald Trump and Vladimir Putin thought they were sharing a drink with Kim Jong Un… but Kim had a little trick up his sleeve! 😂🥤 Made with AI, just for fun. #KimJongUn #DonaldTrump #VladimirPutin #Funny

73 578 4K 182K 903

View Details

Tuan @Tuankxae

7 hours ago

@appltrack $299 is more than double what the last one launched at. The A19 is basically the price of admission for Siri AI. Worth it only if Siri AI is good on day one, and it's still beta in English in 8 countries.

0 0 5 2K 0

View Details

Tuan @Tuankxae

9 hours ago

@YxzRainy 没有cursor原来的套餐好，坚决不转套餐

0 0 0 126 0

View Details

Tuan @Tuankxae

9 hours ago

@goan999999 x.com/tuankxae/statu…我就是这么玩的，不过用的是bot的云电脑，dmit你要买4g以上的内存才行

Tuan @Tuankxae

a day ago

I connected cloud Claude to my Mac so it can work with local files, take screenshots, click, type, and use my browsers. Here’s the beginner walkthrough: SSH + MCP, desktop permissions, Computer Use, backup routes, and recovering the same conversation after a disconnect.

4 0 6 1K 0

0 0 0 560 1

View Details

Tuan @Tuankxae

9 hours ago

Under the new X Subscription, Claude and GPT in Cursor aren't included: they're billed at API list prices from the first use. My read: if Grok isn't clearly ahead, making rival models cost extra is an unwise move. Existing Cursor users: I'd keep the old plan. Switching is optional—and irreversible.