daniel_mac8's profile picture. Agentic Engineering @ampcode | Writing Token Stream | Goodness, Truth and AI | Building at http://github.com/DannyMac180

Dan Mac

@daniel_mac8

Agentic Engineering @ampcode | Writing Token Stream | Goodness, Truth and AI | Building at http://github.com/DannyMac180

ปักหมุด

Gemini 3 Deep Think asked for 1 truly novel human insight. Whoa. 🤯 Warning: You may not be able to sleep tonight. h/t @adonis_singh for the prompt.

daniel_mac8's tweet image. Gemini 3 Deep Think asked for 1 truly novel human insight.

Whoa. 🤯

Warning: You may not be able to sleep tonight.

h/t @adonis_singh for the prompt.

it might not be a bubble.

not enough people are emotionally prepared for if it’s not a bubble



“Direct access to 𝕏 data” is interesting if you take posting on here seriously.

Grok 4.1 Fast just released with our new Agent Tools API that has direct access to 𝕏 data, web browsing and code execution x.ai/news/grok-4-1-…



Dan Mac รีโพสต์แล้ว

METR (50% accuracy): GPT-5.1-Codex-Max = 2 hours, 42 minutes This is 25 minutes longer than GPT-5.

deredleritt3r's tweet image. METR (50% accuracy):

GPT-5.1-Codex-Max = 2 hours, 42 minutes

This is 25 minutes longer than GPT-5.

PSA: Gemini 3 benchmarks are amazing. You won’t know if the model lives up to those scores or not for at least a few weeks. Even then, you can’t rely on other people’s experience. You have to use it yourself for real work and see how it goes.

daniel_mac8's tweet image. PSA: Gemini 3 benchmarks are amazing.

You won’t know if the model lives up to those scores or not for at least a few weeks.

Even then, you can’t rely on other people’s experience.

You have to use it yourself for real work and see how it goes.

Dan Mac รีโพสต์แล้ว

There’s no revealed preference stronger than what you do when you’re trying to help save your child I can now say from experience that, in a medical emergency, you can & should use AI a lot more, and if you do, you’ll get tremendous value

"Nobody is a decel on the pediatric oncology ward" - @labenz

RokoMijic's tweet image. "Nobody is a decel on the pediatric oncology ward" 

- @labenz


Landed in NYC for AIE CODE Summit and see there’s a new code king: GPT-5.1-Codex-Max 👑 > OAI observed it working for 24hrs > #1 on SWE-Bench Verified at 77.9% Good timing!

daniel_mac8's tweet image. Landed in NYC for AIE CODE Summit and see there’s a new code king:

GPT-5.1-Codex-Max 👑

> OAI observed it working for 24hrs
> #1 on SWE-Bench Verified at 77.9%

Good timing!
daniel_mac8's tweet image. Landed in NYC for AIE CODE Summit and see there’s a new code king:

GPT-5.1-Codex-Max 👑

> OAI observed it working for 24hrs
> #1 on SWE-Bench Verified at 77.9%

Good timing!

Gemini 3 Pro asked for 1 truly novel human insight.

daniel_mac8's tweet image. Gemini 3 Pro asked for 1 truly novel human insight.

Dan Mac รีโพสต์แล้ว

Gemini 3 Deep Think asked for 1 truly novel human insight. Whoa. 🤯 Warning: You may not be able to sleep tonight. h/t @adonis_singh for the prompt.

daniel_mac8's tweet image. Gemini 3 Deep Think asked for 1 truly novel human insight.

Whoa. 🤯

Warning: You may not be able to sleep tonight.

h/t @adonis_singh for the prompt.

Dan Mac รีโพสต์แล้ว

"Gemini 3 is too expensive" is a horrifying take. Do you know how much GPT-3 cost per 1 million output tokens? > $20. Its score on ARC-AGI? > 0%! Gemini 3 is not too expensive. You're too cheap.

daniel_mac8's tweet image. "Gemini 3 is too expensive" is a horrifying take.

Do you know how much GPT-3 cost per 1 million output tokens?

> $20.

Its score on ARC-AGI?

> 0%!

Gemini 3 is not too expensive. You're too cheap.
daniel_mac8's tweet image. "Gemini 3 is too expensive" is a horrifying take.

Do you know how much GPT-3 cost per 1 million output tokens?

> $20.

Its score on ARC-AGI?

> 0%!

Gemini 3 is not too expensive. You're too cheap.

Loading...

Something went wrong.


Something went wrong.